Agent Failure and Market Failure

Steven Horwitz

The Buchanan conference is off to a great start.  One issue that came up this morning is the role of behavioral economics and psychology within political economy.  What's interesting is that it came up in a session also devoted to talking about externalities.

One thing that has always struck me about behavioral economics are the parallels to arguments about market failure (such as externalities).  Much of BE shows us that economic agents are not the rational utility-maximizers of standard theory (as does much of psychology of course).  But what this result implies for economics and policy depends upon whose hands the results are in.

One school of thought seems to say "a-ha!  Agents don't act like our ideal models of behavior, therefore agents fail to behave in the way necessary for those models to work."  Thus, in this view, "agent failure" leads to "market failure."  And notice the parallel structure of the two "failures:"  in both cases, failure is defined as not matching the idealized, perfect result, either perfect rationality or perfect competition/general equilibrium.  The remedy, of course, is some combination of paternalism and intervention to bring the failed agent or market closer to the modeled ideal.

In the hands of some others, especially those working in the Vernon Smith tradition, the "agent failure" discovered by BE becomes not a cause to castigate the real world for not matching the model, but an opportunity to explore how imperfect agents ever in fact learn sufficiently well to generate outcomes that are still not perfect, but are much closer to rational than any alternative.  In Vernon's "ecological rationality," it is institutions that do, to use Pete's phrase, the "heavy lifting" of generating social order, not the behavioral assumptions we make about agents.  In other words, the imperfections of agents are an opportunity to use market institutions to achieve outcomes comparatively superior to any alternative set of institutions (though those results are still "imperfect").  Rather than condemning agents for their imperfections ("failure"), we should ask how order still gets generated despite those imperfect agents and whether that order is better or worse than alternatives.

Notice how this too parallels the market failure literature.  The response by various approaches (e.g. UCLA, Virginia, Austrians, NIE,) to Samuelsonian externality arguments has been to see such supposed market failures not as cause for condemnation but as an opportunity to do one of two things:

1.  Engage in comparative institutional analysis and ask whether "failed"  markets are still better than "failed" government intervention

2. View market "failures" as opportunities for entrepreneurs to step in with ways of solving the externality problem and reducing the relative imperfection of the market outcome.

Both solutions are not only possible, they are compatible:  part of what makes imperfect markets better is that welfare-enhancing entrepreneurial action is more likely there than in politics.  

It would also be interesting to see whether the move to BE in recent years is the result of market failure arguments failing to be as persuasive in light of decades of criticisms of the sort noted above.  If one wants to be more cynical, one might see the failure of market failure theories (there's a great paper title!) as the motivating force to turn away from markets and toward agents as the element that fails to match the model and thereby brings down, in the eyes of the market failure crowd, the argument for the market.

Part of the problem here is language:  we need to jettison the language of "market failure" and perhaps replace it with "market imperfections."  At least the latter suggests there is still a functioning market there capable of correcting its imperfections.  A "failed" market would not seem to be capable of self-correction.   Or to use some language Dick Wagner has used:  an "imperfect" market still is capable of "becoming," while a failed market is just "being" failed.

Similarly, we should be sure to avoid using the results of behavioral economics to talk about the equivalent of "agent failure."  As we've known since Menger, human beings are imperfect actors, caught between alluring hopes and haunting fears and stumbling and bumbling our way through an uncertain world.  We  "fail" all the time and it is because of the institutions of the market such as property rights, contracts, prices and profit/loss, and the possibility of economic calculation that they bring, that we are able to overcome our limits and produce the order that we do. 

Where externality theory sees market failure and behavioral economics sees agent failure, Austrians and others see the normal imperfections of human life and the opportunity to understand how the rules and institutions of the market are resilient enough to enable us to overcome those imperfections and get Paris fed.

Comments

[Daniel Kuehn]

Very interesting post.

I would tend to agree with you on the agent failures but not on the market failures. The reason why markets are so successful is that the market process is robust to these agent failures. That's precisely the point.

And certainly any government intending to intervene is going to be subject to precisely the same agent failures as market participants, so paternalism doesn't really offer a solution at all if the goal is some sort of welfare maximization or optimality (if the goal is, for example, to get people to open 401k's then certainly paternalism informed by BE can help you reach that goal - but getting people to open 401k's is a very different goal from maximizing welfare!).

I'm not sure how this connects to market failures, though (at least conceived as an externality). If we think of the market process as a process that takes imperfect agents and their information, and their incentive to act on that information (I'm not sure if you all would phrase it differently), the problem with a genuine market failure as I see it is that something is missing - either the information or more damningly the incentive to act on that information (ie - an incomplete property rights regime). The market can potentially get around "failures" like information asymmetries and things like that (but then again, it might not and there might be a relatively costless way for the state to help get around it). But when it comes to actual externalized costs or benefits because of the nature of the property rights regime, I don't see how a problem outside the market process like that can be self-corrected by the market. They seem to be two fundamentally different things - one offers the opportunity for entreprenuerial agents to make better use of imperfectly wielded resources, information, etc. - the other is the complete absence of an incentive.

Now - I personally view the state as an emergent instutition just like the market or the firm or an Ostromesque norm arrangement, so I personally don't see a fundamental contradiction between the idea of emergent market processes to solve problems and a state (which emerges and evolves over time to do these sorts of things) that solves the problem. If there is an externality, though, I would expect that if anyone is going to solve the problem it's going to be emergent behavior in the government, rather than emergent behavior in the market - because emergent behavior in the market responds to costs and benefits, and the whole problem with an externalities is that the costs and benefits aren't there to act on!

So good thoughts - but these two still seem different.

I would agree that things like asymmetric information, etc., aren't quite in the same class of problematic market failures as externalities.

[Daniel Kuehn]

*and I totally agree on jettisoning bad language point. "Market failure" is a habit, but it's habit I'm more than happy to try and break because I agree "market imperfection" is better. Or better yet - just call it what it is. Is it an information asymmetry? Is it an externalized cost? Just call it that!

[Troy Camplin]

We should rather say "ideal market" and "real market". There is only market failure qua markets insofar as those markets do not match up with the unrealizable expectations of ideal markets. It is much like saying that there are no real elephants because no actual elephant meets the criteria of an ideal elephant. Thus, all elephants fail to be elephants, and there is elephant failure! Seen this way, we can see just how absurd the idea of "market failure" really is. Thus, it's not even "market imperfection". There is no imperfection except insofar as it doesn't mean an unrealizable ideal. It is imperfect compared to perfect, but utopia is not an option.

I will say that you will notice how almost noone says in regards to agent imperfection, "Aha! And that's why we shouldn't fool ourselves into believing we can control or even fix things with government -- they too are fools, just bigger ones for thinking they know more than the rest of us!"

[Daniel Kuehn]

Troy -
Another way to look at it is that given social institutions (elephant anatomy) there's some optimal state constrained by technology, institutional structure, etc. that we expect society (an elephant) to be able to reach. Now, any given case (every given elephant) is going to be littered with various and sundry imperfections (genetic variation, bad upbringing, chance accidents, etc.). So ya, talking about an "ideal market" ("ideal elephant") is a little silly and besides the point. But saying that a specific failure of a market (a specific failure of an elephant), say an externality (elephant flu... or whatever) keeps society from reaching it's full potential, aside from the natural variations and imperfections.

Usually emergent processes (biological evolution) provide a solution. Sometimes it's a market process, sometimes it's a non-market process (i.e. - either private or public non-market collective action outside of market action).

[Jerry O'Driscoll]

Alchian's take was that our theory is rational, not necessarily the agents whose behavior it describes.

For Mises, rationality was a purely formal concept: human actors apply means to ends.

[Ashwin]

Even the notion that human beings are "irrational" does not square with a lot of the recent work in the neural and cognitive sciences - Gerd Gigerenzer's work shows convincingly that most heuristics/biases are not a sign of irrationality but an entirely rational form of decision-making when faced with irreducible uncertainty. I can recommend anything written by him but especially his aptly named book “Rationality for Mortals: How People Cope with Uncertainty”.

Secondly, for micro-irrationality to add up to macro-irrationality, agents need to be irrational in the same way at the same time.

And lastly, it ignores behavioral heterogeneity and the role of institutions and incentives in selecting from such a heterogeneous pool e.g. to claim that traders in an investment bank are typical in their risk-taking preferences is ludicrous. The key here is that the environment selects for risk-lovers in many cases.

I've written this up in more detail here http://www.macroresilience.com/2010/06/24/agent-irrationality-and-macroeconomics/ . And here is a description of how some "irrational" heuristics outperform other "rational" strategies in asset allocation and how even Harry Markowitz used such a heuristic http://www.macroresilience.com/2010/07/08/heuristics-and-robustness-in-asset-allocation/ .

[Jerry O'Driscoll]

To add to Ashwin's astute observations: the decision that matters for economic theory is that taken at the margin.

[Will Luther]

"We "fail" all the time and it is because of the institutions of the market such as property rights, contracts, prices and profit/loss, and the possibility of economic calculation that they bring, that we are able to overcome our limits and produce the order that we do."

Steve, do you also accept Hayek's Sensory order argument? I take it to mean that we are actually wiser in our context than Cartesian rationality would allow because we have evolved as humans in this specific context with those strategies most fitting for survival/reproduction (things like "trust" in certain Prisoners' Dilemma games) winning out.

[Bill Stepp]

The notion of "market failure" anthropomorphizes markets and should be dropped IMO. Only entrepreneurs fail or succeed by earning profits or suffering losses.
The "emergent state" emerged as a criminal enterprise, a protection racket; the modern state is a more massive criminal enterprise that makes Enron and Bernie Madoff look like choir boys.
Mises notion of "rationality," i.e. actors applying means to attain ends, is as far as that sterile concept should go. The notion of irrationality should be ditched.

I assume that Daniel Kuehn's "public non-market collective action outside of market action" is a baffling term for state action. Why not just say "state action"?
See Franz Oppenheimer and Jonathan Haas on the origin of the State.

[Ryan Murphy]

Professor Horwitz, can you provide more context about how externalities fit into Behavioral Economics? To my knowledge, most of the things that Ariely or Kahneman or Thaler point to are apparent violations WARP or some other variation on neoclassical rationality. The losses are internalized to the agents; they just don't respond "correctly." The only person I can think of that would relate to coordination failure and externalities would be someone like Robert Shiller, and he isn't exactly the first name you think of when you say "behavioral economics." In general, the behavioral economic thing is probably better thought of as an internal equilibrium failure rather than a failure of the market process.

[The Cuttlefish of Cthulu]

"I assume that Daniel Kuehn's "public non-market collective action outside of market action" is a baffling term for state action. Why not just say "state action"?"

Excellent point.

[Greg Ransom]

Is the BE and/or market failure approach ever applied to ... academia?

Or the output of "economic science"?

Just asking.

[Greg Ransom]

More specifically, do BE theorists very attempt to apply their framework to themselves and their own choices and output as "scientists"?

Or do market failure theorists look at the opportunities for correcting market failure in the market for scientific ideas in economics?

[Steve Horwitz]

"I assume that Daniel Kuehn's "public non-market collective action outside of market action" is a baffling term for state action. Why not just say "state action"?"

Because there is a lot of "public non-market collective action" that isn't necessarily "the state." Communities can form all kinds of voluntary systems of rules that are public but not market, or political but not "the state." I suggest reading some Lin Ostrom on this point, not to mention a number of other folks.

Market and state do not exhaust the forums in which rules of governance can be developed.

[Steve Horwitz]

@Ryan: I wasn't arguing that externalities "fit" into BE. I was simply arguing that there's an analogy in the style of argument between BE views of "agent failure" and the theory of "market failure," which was discussed today in the context of externalities.

[Steve Horwitz]

@Will: Yes. :)

[The Cuttlefish of Cthulu]

I'm not sure I understand your response to Bill Stepp, Steve. Isn't the "market" just a short-hand for the nexus of *voluntary* interactions, that are categorically different from the nexus of coerced interactions, ie the "state?"

[Troy Camplin]

The market is a subcategory of volutnary interactions, but it doesn't exhaust them.

Daniel,

The problem is that one cannot know what a market's "full potential" could possibly be. If I knew what the market's full potential could be, I would provide the services, technolgoy, institutions, etc. that would bring it to its full potential and become a billionaire. So would anyone else. So while it is true that no market is ever going to be optimal, or at its "full potential" (whatever that may mean -- I suspect that a truly free market really is at or approaching its full potential at any given moment, though, of course, I cannot know that any more than anyone else could), that doesn't necessarily argue for any particular intervention into the system. Also, such optimality begs the question, "Optimal for what?" Which then makes it clear that we are talking teleology; and once we are talking goals, we are no longer talking about a market economy and what it actually does. Hayek made this abundantly clear. A crocodile, for example, is not an optimal companion pet (notwithstanding that my 3 yr old daughter would disagree with me on this), but it is well-suited for what it evolved to do.

[David Andersson]

Interestingly, I have found it more difficult to find compelling (sustained and deep) discussions about public goods/externalities in the Austrian literature (but perhaps I'm just ignorant). On the other hand, economic theories of property rights are particularly strong at addressing the narrow-minded view of conventional market-failure-based interventionism. Alchian, Demsetz and Barzel are very good at this. And then there is also an outstanding discussion of market failures that bring in a spatial element in the Calculus of Consent. In addition, there are more explicitly spatial property rights approaches such as those of Wagner (GMU), Foldvary, Webster and Lai etc. What am I trying to say here? Basically, that any discussion of public goods/extenalities (Buchanan thinks this is one and the same thing, and I agree) or "market failure" generally benefits from being addressed in terms of spatial property rights theories (i.e. institutions/property rights/land values/accessibility to spatial attributes with high exclusion costs).

[RPLong]

I agree with Bill Stepp here. Just because an economic agent's actions don't lead to the best calculable result in hindsight doesn't mean the agent was irrational. It's impossible to calculate every economic activity one engages in. Sometimes the time saved avoiding the calculation is "worth it" to the agent. It doesn't mean the agent failed or didn't maximize utility, it just means that the economist hasn't considered all the relevant factors in his/her model.

If a photo looks fuzzy, you might have to focus the lens, zoom in and out a little, until you have a clearer picture of what's going on. Mises would call it "vain" to assume that our best economic models give a perfect picture of what's going on economically, even in simple decisions.

[Bill Stepp]

Steve/Troy,

Yes, there are other sorts of voluntary institutions, such as community associations, etc., but they are categorically different from the criminal entity known as the State and belong in the same category as the market for my purpose.
Anyway, if we broaden the notion of profit/loss to psychic profit/loss, the distinction collapses, and we are left with voluntary institutions and criminal entities.

[Roger Koppl]

@ Will & Steve,

I think it's a bit of an overstatement to say "we have evolved as humans in this specific context." We evolved in the Pleistocene as hunter-gatherers living in small bands of 15 -150 persons. We live in A. Smith's "great society," which is a different context. That difference in context gives you V. Smith's distinction between personal and impersonal exchange. I think that's pretty important for sorting out the different roles of different social institutions, how and why things can wrong in the great society, and so on. I think Hayek pretty much had the point, as in his "Three Sources of Value" and "The Atavism of Social Justice." But Cosmides and Tooby add greatly to our understanding of the difference in context and why it matters.

[Fearsome Tycoon]

I always thought that what Mises meant by "rational action" was simply that human actions have a purpose, not that they are perfectly logical and always make the best decisions. And wouldn't a market with no "failures" be a market where no one ever took a loss on anything, since all ventures would maximize the utility of the inputs?

The difference between markets and interventionism is that markets don't allow you to do useless or foolish things for very long. Even if you never figure out what you're doing wrong, you will simply run out of money and be forced to quit. All interventionism accomplishes is propping up lossy ventures much longer than they would be able to live on their own (ethanol is a great example of this).

[Daniel Kuehn]

@Cuttlefish and @Steve -

On non-market emergent solutions, that was precisely my point that there are both state and non-state institutional solutions that are non-market. But in addition to that (and this is likely where Steve might start to diverge from me), I think it's important to highlight that state action is emergent action. All human institutions evolve in response to their environment and the sorts of problems that people want to solve. Some state solutions are going to be inappropriately centralized and they may suffer from that, but I think it's wrong to ontologically ostracize the state the way a lot of people do. That sort of thing doesn't seem to me to have any scientific basis. The other reason to class the state with the sort of non-market governance that Ostrom works with is that I think in all likelihood a lot of those non-market governance structures are what evolve into states. Trade leagues, committees of correspondence, etc. - these often evolve into state structures. And even once you have a state, the state evolves over time. So your "state action" isn't inaccurate, but in this case - talking about solutions to imperfections - I think that it confuses thing to think of the state in a class by itself.

[Steve Horwitz]

@Bill:

I think, following Mises, that there is a difference between voluntary organizations that operate in the realm of catallactics/monetary exchange and those that operate in other realms and by other principles. My family is neither the market nor the state, for example. Usenet groups are neither the market nor the state. Both are voluntary organizations but both have explicit and implicit rules created/discovered by their participants in broadly political processes.

If you want to distinguish voluntary from coercive organizations, that's perfectly fine. My point is that those don't map one-to-one to "market" and "state." There are other non-market (i.e., not catallactic) voluntary organizations.

[Garrett Watson]

I have to agree with Bill Stepp above. Mises clearly explains in Human Action that the very notion of an 'irrational' action is flawed. While people may not achieve the results desired in any given action, that fact alone does not make their action 'irrational'.

As for 'market failure', I've always found the doctrine lacking an acknowledgment of market corrections to the supposed "failure". "Pure and perfect competition" specifically is especially flawed, and pretty much is as far outside of the true nature of competition as one can get, IMO.

[Roger Koppl]

I admit I have not been following the comments closely, but I would like to comment on the rationality assumption. If all action is rational, what use are we getting out of the word "rational"? I'm just cribbing from Fritz Machlup. I think he was right. It's not so much that Mises was somehow wrong in his definition of rational. It is rather that using the perfectly serviceable word "rational" so broadly effectively strips it of any meaningful use. As Mises himself noted, the phrase "rational action" is "pleonasm." I'd rather give a more narrow meaning so that the word might, sometimes at least, do some work for me.

[Jerry O'Driscoll]

I have no disagreemnt with Roger.

But orthodox economics packs a great deal into rationality. So much so, it invites rejection on any number of grounds. That rejection then leads some to conclude that all of standard theory is suspect.

When we get to rational expectations, a new set of problems. In its most basic form, Ratex states that individuals will make the best use of available information. What are we entitled to assume inviduals know, and how should they make use of what they know? If they fail to follow the model's predictions, are they irrational? Or is the model?

[Fearsome Tycoon]

Roger, you have to interpret Mises against his own background, where the dominent intellectual figure was Karl Marx. According to Marx, human action isn't rational; it's merely caused by material conditions, much like the orbits of the planets or lightning striking the earth. In a Marxist framework, talking about the "reason" for an actor's action is useless; you need to focus on his class identity and material conditions instead.

The fact that "rational action" seems like a useless category today speaks volumes about how irrelevant Marx is to modern economics.

[Barkley Rosser]

On the opening day, which I attended, Jim Buchanan criticized rational expectations.

Sorry Pete closed comments on his opener about the conference. The first day was interesting and there would have been the place to make comments on it. Pete did a good job opening everything up.

[Roger Koppl]

Jerry: Yes, absolutely.

Fearless: I think the context was Max Weber, not K. Marx. Mises was saying all action fits MW's ideal type of instrumentally rational action. He was right, IMHO, but his substantive point may not have required that we gut "rationality" of it's discriminatory power.

[Jerry O'Driscoll]

Barkley,

What did Buchanan have to say about rational expectations? He was never happy with it.

[Mario Rizzo]

The problem with the Misesian approach to rationality is that it annuls the standard welfare economics. If preferences are only knowable as inferences from actual choices, then there can never be any deviation of choice and preference. This is not the case by any argument or brilliant insight but by analytical construction. All of the decisionmaking failures -- like lack of willpower -- discovered by modern psychology and by behavioral economics are dismissed a priori. Not a good strategy today.

(I should add that Rothbard unfortunately tried to get a welfare conclusion from the Misesian rationality approach. Mises himself, however, did not try to do this. Mises's welfare economics, if you can call it that, was simply to say that interventionist policies did not produce the results its advocates proclaimed.)

My preferred approach is to think of rationality, even at the individual level, as an equilibrium phenomenon. This means that the ex ante equilibrium of an individual's plans is not to be treated as a tautology.

I do not believe that one will always be able to show that agent rationality doesn't matter very much and the appropriate institutions will always come to the rescue.

Therefore, I believe that we must continue to focus, though not exclusively, on agent rationality.

No deus ex machina in the market.

[Barkley Rosser]

Jerry,

If you go back to the post that Pete Boettke put up on Buchanan's address to the history of thought conference in Richmond back in June on the "Old and New Chicago," you will get the idea. He basically summarized that argument and specifically cited that paper. He does not like the "new Chicago" of Fama and Lucas, with their ratex assumptions. He prefers the "old Chicago" of Knight and Simons. Said that markets need a proper institutional and legal framework to operate properly. Thinks that some of the more complicated financial derivatives should have been banned.

[Troy Camplin]

The problem is that there are at least three definitions of "rationality". Hayek identifies two of them -- Cartesian/constructivist reason and Scottish reason. However, as has been noted here, there is also the economists' reason, which I found to be brilliantly explicated in an article in a recent issue of the journal Biological Theory that asked if plants were rational. Well, according to the standard economists' defintion of rationality, even plants are rational. There are good reasons why they do things, such a turning toward light, or in the case of some vines, wrapping around twigs once they come in contact with them. A cat is rational: the reason it lies in the sun is to get warm. The problems arise when there is confusion as to which reason is being considered. We have one name for three different concepts.

[Daniel Kuehn]

@Troy -

- Do you think a Biological Theory article trying to say "economists think plants are rational" is the best place to get a reasonable view of how economists think about rationality? It might be but I would have gone other places first.

- I think "economist's view of rationality" is largely just a modeling assumption - it's a way to make it tractable. It does reflect a lot of constructivist thinking, but it's not dependent on it in the deep form that Hayek explains constructivism. Economist's rationality is essentially that, observing costs and benefits, agents will be able to weigh them soundly and choose the greatest net benefit. That's what it essentially boils down to, right? That seems to be a basic tractability assumption for modeling that is quite easily integrated with either constructivism or Scottish reason because - as Roger Koppl said earlier (of Misesian rationality) - it doesn't really ask much.

[Daniel Kuehn]

And I'm not even saying that that vision of rationality always holds true - simply that you don't need to be some sort of constructivist human computer to satisfy that definition.

[Mario Rizzo]

Milton Friedman talked about the "rationality" of plants moving toward the sun, etc. What is not sufficiently appreciated is what this "as-if" approach does to welfare economics. It means that the "subjectivist" standard no longer makes sense. The plant's welfare must be defined in terms of what *actually* benefits it. The standard is "agent" relative but objective.

Now if human agents are rational in the same way then the relevant welfare criterion will be an objective (but agent-relative) one.

These implications are serious and that is why I am not willing to give up on actual agent-rationality!

[Daniel Kuehn]

Mario, re:"What is not sufficiently appreciated is what this "as-if" approach does to welfare economics. It means that the "subjectivist" standard no longer makes sense. The plant's welfare must be defined in terms of what *actually* benefits it. The standard is "agent" relative but objective."

Could you explain why every economist that is always alleged to be a closet objectivist expresses what is good in terms of subjective understandings of utility?

I don't know what Friedman said about plants - but everything I've ever heard anyone calling themselves an economist has said about humans gets back to subjective preferences driving choice.

Is your understanding different? I'm with Bryan Caplan on this - it seems like every time we say "I'm subjective too - I don't even understand how objectivism makes sense for understanding choice", the response is always "well we're RADICALLY subjective" with no real elaboration of what difference that implies.

I suppose I'm just curious because your assertion of objectivism doesn't seem to be reflected in any of the economics I've ever learned.

[Jerry O'Driscoll]

@Barkley,

Thanks.

[Barkley Rosser]

Jerry,

You're welcome, and thanks to Pete for getting the Thursday session up here on video.

[Roger Koppl]

Daniel Kuehn asks Troy caustically whether an article in Biological Theory "is the best place to get a reasonable view of how economists think about rationality?" Um, dude, perhaps you should check out the credentials of the, um, author of said article before, like, copping an attitude?

[The Cuttlefish of Cthulu]

"I suppose I'm just curious because your assertion of objectivism doesn't seem to be reflected in any of the economics I've ever learned."

I agree with Daniel Kuehn on this one. Too much Austrian work on this topic tries to create a distinctiveness where none really exists.

[Mario Rizzo]

The dirty little secret is that modern welfare economics is incoherent.

If agents need not be actually "rational" then it makes so sense to say that they have subjective preferences. Consider if a person is like a plant then the only sense in which *it* could be better off is objectively, that is, in terms of what will promote its existence, its size or whatever the analyst is concerned about.

To view this kind of agent as having subjective preferences is prima facie nonsense.

But if that doesn't satisfy you, then consider this. If we adopt the Misesian tautological perspective (as he himself calls it) then we infer (subjective) preferences from actual choices. So whatever the agent does is rational. So what does it gain us to call it "rational"? The agent just does what it does. What does it gain us to say that it furthers its subjectively perceived welfare. It always does!

And it does all of this just by how we set things up -- how we defined our terms. WE HAVE SAID NOTHING SUBSTANTIVE ABOUT THE WORLD.

Now if we argue as Friedman did that we can say the agent acts "as if" it were rational because the theories generated lead to good predictions, we have not said anything about subjective welfare.

If we argue as does Vernon Smith that agents need not be rational for the market to be "rational" we still have said nothing directly about subjective welfare.

It seems to me the only way we can get at welfare in these two approaches is to define it in terms of what the analyst believes is best for the agents. (Enter the behavioral economists.)

Now I wish to reject all this. To do so, I must affirm the importance of dealing with individual agent rationality in a non-tautological way. I cannot abandon the issue and say institutions, such as the market, will make things A-okay.

I guess if you are willing to give up normative welfare economics completely you could go this route. Economists like Gul and Pesendorfer have done this. I have been tempted to do this too.

[Roger Koppl]

I'm not sure I quite see the problem, Mario. What do you want "normative welfare economics" to do?

[Bill Stepp]

Didn't Rothbard write a good Austrian critique of welfare economics?
Whatever an agent does is rational, and the concept of an irrational act has no place in economics. Sort of like saying A is A. That doesn't mean a tautology is useless.
The tautology that "man acts" doesn't say much, until someone denies it, as George Selgin use to say.
It's unfair to Mises though not to go beyond this. You have to take the next step, and go beyond this by considering an individual agent, his value scales, choices, and an exchange--action, in a word.
I see nothing in the discussion so far about action. The day Austrians stop talking about action, they might as well sign up for the school of Samuelson.
As I said before, the concept of a rational agent is sterile. That doesn't mean it's useless. You want a bandaid for a wound to be sterile, after all.

[Mario Rizzo]

Traditionally normative welfare economics is supposed to evaluate states of affairs in terms of the preferences of the individual agents themselves.

[Roger Koppl]

Mario,

Sure, of course. Hm . . . Maybe I'm totally not tracking. You said you are "tempted" to give up "normative welfare economics." I took you to mean that you're not really there, at least not yet. So you see value in something you are calling "normative welfare economics." Your last comment seems to say that the version you are still trying to hang onto is pretty much the tradition version of, I guess, post-war microeconomics. Now I never felt like I got much benefit from concepts such as HIcks-Kaldor efficiency or even, frankly, consumer surplus. Maybe that overstates a bit. For example, in experimental studies where you want to know if the surplus is extracted. But I just don't see the big opportunity cost if we have to run our normative analysis in a different way. On the contrary, static vs. dynamic efficiency and all that. So, again, what is it you want (traditional?) welfare economics to do? What's the opportunity cost of pitching it out the window?

(I guess you will reply with something about BE and paternalism, but I would tend to think your better off answering paternalists without welfare economics. Anyway, I'd better wait for your answer before replying to it!)

[Mario Rizzo]

Roger,

Maybe you never got much out of the standard welfare economics. Nevertheless, it has had a significant role in economics.

You are right to suggest that my main interest is in paternalism and behavioral economics. And yet the issues raised here are ubiquitous. Can we infer from the agent's actions that he is furthering his welfare as he see it?

I say no unless there is the possibility that he might not be doing so. We need something more than a tautology here.

I want to say that only in ex ante plan equilibrium do the individual's choices promote welfare as the agent see it. And then I want to discuss whether there are empirical movements toward that equilibrium.

This is not Friedman's or Vernon Smith's approach. They say, "Don't worry about individual rationality; markets will exhibit the outcomes those who postulate individual rationality seek even if agents are irrational."

All I am saying is that the BE claim that agents do *not* always advance their welfare, as they see it, is correct. Or at least there must be the *possibility* that it is correct if anything resembling normative welfare economics is to make sense.

Otherwise, we must simply say that economics has nothing to say about ranking outcomes from the point of view of the agents themselves.

Do the Vernon-Smithians understand this?

[Troy Camplin]

"Are Plants Rational?" by Economist Elias Khalil, who says, "The actons of plants are rational insofar as plants react to incentives, i.e., they exhibit behavioral flexibility in reponse to a change in constraints. Plants react to incentives when they change their orientation if re-planted, react to changes of soil moisture,... or adjust behavior when exposed to new predators" (Biological Theory 5(1) 2010, 56).

Plants have very few, if any, subjective preferences. The need sunlight, water, and protection from predators, though different species need different amounts of these. Humans need food, water, and protection, though different people want different kinds of these. Plus we have innumerable other wants. It is in the fulfillment of our wants where subjective choice and values comes in.

The point of the article is that standard economists use the term "rational" in such a way that one can talk about plants as rational.

[Daniel Kuehn]

Thanks for the response, Mario -

- I certainly don't disagree with you that a plant can't have subjective preferences.

- I also agree strongly that institutions won't always provide a solution - your deus ex machina point. I think we've evolved some quite robust institutions, though - there's reason to be optimistic without assuming perfection.

- Maybe I'm not understanding your point, but you still seem to be listing a bunch of views on rationality, and then say "but we're still not saying anything about subjective welfare". Well, yes - concepts of rationality can be applied to objectivism or subjectivism, so just knowing what these guys think of rationality won't get you very far. So I would think you would push a little deeper and ask "what did they think of preferences?" And universally it's subjective preferences, just as in Austrian economics.

Roger - I'm sorry you took me to be caustic - I certainly wasn't intending to be.

[Othyem]

Does anyone know of a good introductory BE book that's as balanced as possible?

[Roger Koppl]

Mario,

I guess I still don't quite see the problem. I did *not* detect in your reply any claim that we cannot criticize libertarian paternalism without welfare economics. (And I would not have agreed with such a claim.) Rather, if I'm tracking, you want to be able to ask which arrangements tend more to generate "outcomes" favorable "from the point of view of the agents themselves." And you wish to ask that question quite apart from any criticism of libertarian paternalism. (My apologies if I'm misinterpreting you.) But the individual point of view is shaped by the very alternative arrangements we'd like to rank. You know, the average man will rise to the level of a Goethe and all that. The preference map is endogenous. So I just don't see where traditional welfare economics, with its exogenous preferences, gets you that far.

I would tend to say go straight to good old fashioned (indirect) utilitarian arguments. You get most of what you want just with Mises' notion that most illiberal policies do not achieve the ends putatively aimed at. For the little that remains to argue about, we have the utilitarian notion of "autonomy" to do most (all?) of the work that might be done by pointing to "the view of the agents themselves." Why pine for the loss of a normative system that assumes asocial agents, exogenous preferences, and a static world. (And how does that fit with Bergsonian time?)

[Pietro M.]

Just a note/question on Mises's rationality assumption.

In no place in Mises's writings I got the impression that he used the term with the same meaning as "rational agent" in standard economics.

"Rationality" in Mises was a critique of the irrationsists, those who didn't believe that a science called economic theory could be made because men were irrational.

"Rational" in Mises means that economists and the historians can investigate it, it doesn't imply anything regarding the actual behavior of economic agents and/or the successfulness of their endeavors.

From this point of view, of course, Mises's definition was devoid of content, purely tautological, with no additional insights.

[Rafe Champion]

If you really want to get confused about rationality, check out Popper's Rationality Principle RP, the driver of actions in his Situational Analysis, which he variously described as essential, almost always false, and possibly empty! More situational analysis required? (Actually it is his take on the "logic of pure choice).

Reason and rationality are among the words like liberal, conservative and induction which have so many different meanings that you sometimes wonder if discussion would go better without using the words at all, just using more words to say what you mean in in each situation. Rogers Brubaker in "The Limits of Rationality" noted sixteen apparent meanings of "rational" in Weber.

[Jonathan]

A note to Greg Ranson. See Daniel Ariely's Predictably Irrational for examples of introspection on his own experience as a motivator to research. Dick Thaler also "took notes" on the behavior of economists in formulating ideas about mental accounting.

Of course the failure of Leonard Savage to comply with his own axioms is well known (Allais Paradox).

[Kyle8]

It seems to me that if agents do not always act as the "perfect" agents of theory, That, in itself is not a problem.

The real problem is, does your theory take actual human behavior into account in a better way than a competing theory.

For instance. A theory of free markets and laissez Faire does not have to prove that every person will maximize their potential in the way that perfect theory predicts. It is only necessary to show that it much more accurately predicts the behavior of free agents than a competing theory of let us say Keynesian government control..

[Troy Camplin]

One could even argue that it is the system that best allows people to maximize their potential.