Jungles, matches & optimality

Amidst all the celebration of Lloyd Shapley and Alvin Roth's Nobel, there's something nagging at me.

Their achievement was to show, theoretically and empirically (pdf), that we don't necessarily need a price mechanism to produce an allocation that is stable and Pareto optimal. Another process can do so, which works for matching students to schools, kidneys to recipients and for marriages, among other things.

But there is another mechanism that also achieves such a result. It's Ariel Rubinstein's jungle (pdf) economy. He shows that, under certain conditions, allocating scarce resources to the strongest agents according to their preferences will also give a stable, Pareto optimal equilibrium.

And, of course - in theory and under some conditions - a free market will also achieve this.

We have, then, not one or two but three ways of getting to a stable Pareto efficient outcome.

However, most of us, I suspect, don't think Rubinstein's jungle economy an especially pleasant place to be - even though it achieves the same desiderata as the (theoretical?) free market and Shapley-Roth process: stability and Pareto optimality.

Which raises the question: what do the latter have to make them desireable that the jungle economy lacks?

I don't think economic growth is the answer: ceteris paribus, I would rather live in a free market economy with zero growth than in a zero-growth jungle economy.

Nor is it that the market and matching economies do away with the power that generates the jungle allocation. One feature of Shapley-Roth processes is that they favour the side doing the choosing. Power thus helps determine allocations, as in the jungle.

Nor is it sufficient to say that the Shapley-Roth process saves lives. It does, but Rubinstein claims that the jungle allocation also does so by removing the need for violence.

And I don't think it's good enough to make the empirical claim that we in fact rarely see Rubinstein's jungle economy working as well as theory implies: the same is true for free market economies.

What I'm edging towards here is a point made (pdf) - in a different context - by Amartya Sen, that Pareto optimality isn't as great an ideal as economists think.If you can't abolish slavery without making slave-owners worse off, then slavery is Pareto optimal. 

To return to the question: why do we cherish the market economy over the jungle economy? I suspect one reason is Deirdre McCloskey's - that a market (sometimes) promotes more attractive virtues than other societies. It is, perhaps, significant that the English have a longer tradition of freeish markets than the Serbians.

And this is where my nag arises. If Pareto optimality is not sufficient to make an allocative mechanism attractive, what else is needed, and do Shapley-Roth processes possess it? 

Comments

[Luis Enrique]

"Pareto optimality isn't as great an ideal as economists think"

this is one of those points that make me wonder if my education in economics is some sort of freakish outlier

as far as I can recall, I was taught all along that Pareto optimality is a very weak concept of optimality, and really nothing to do with what we think of optimality in any every day sense of the word, nor anything to do with welfare optimality in a formal economics style either.

Of course Pareto optimality is of interest, because if a situation is not Pareto optimal then there are improvements to be made even without having to think about potentially more complex questions like distributional justice. This is worth knowing.

Obviously, very many terrible situations are Pareto optimal because, for example, you cannot make the exploited better off without making the exploiter worse off. This is a trivial point, and I would say an example of people's lamentable tendency to think that those stupid economists haven't realised something that should occur to anybody after about one minute's thought ... except that you seem to be saying here that economists regard Pareto optimality more highly than they ought.

Cosma Shalizi thinks so too:

http://masi.cscs.lsa.umich.edu/~crshalizi/weblog/942.html

[Metatone]

Luis - where did you get your education?

My impression is very much that it's a kind of outlier.

Certainly almost any mainstream economics dept. in the USA looks a lot like Shalizi's outline.

[Metatone]

My experience in the UK is that many UK academic general economists are much more nuanced about optimality. However the economists who colonise other fields (e.g. my field, health organisation, so I'm talking about health economists) tend to be utterly rabid in defining outcomes around Pareto optimality. Not sure why that is... but it's definitely noticeable.

[Luis Enrique]

I'm pretty sure that my very first micro lecturer told me that, this guy:
http://www.ems.bbk.ac.uk/faculty/wright/

[Luis Enrique]

Metatone,

well that's interesting ... I have a little experience with health economics, just one module on my Msc, and I can remember some work based on welfare maximization that had some discussion of the fact a social welfare function might care about distributional questions, if you see what I mean, but those questions were going to be set aside for purpose of analysis.

p.s. I just checked Wright's Birkbeck lecture slides online and found this:

But perfectly competitive framework provides the optimal outcome in the (limited) sense that it is “Pareto-Efficient”.

which is consistent with what I remember, long time ago now.

[Alex]

My experience is that intro economics will indeed tell you that Pareto optimality is actually not all that great, but then the rest of economics will constantly harp on it and wave it at its critics every five minutes. Similarly, intro economics usually tells you that the standard assumptions in the theory of the firm in perfect competition are hilariously unrealistic, but the rest of economics then ignores them as far as possible.

[Alex]

On another thread regarding the jungle economics paper, I pointed out that the Lange model result suggests that you can have a Pareto-optimal outcome in a planned economy. Perhaps any economic system is capable of arriving at a Pareto outcome?

[Metatone]

@Alex - isn't it a sort of definitional problem? By the time you've excluded anything that doesn't fit into the definition "economy" you've basically fulfilled Pareto?

[Keith]

I am not sure serbian conduct is explainable by economics but by Psychology. Behaving in a offensive anti social manner and being unable to admit it and apologise is a form of aggressive denial. No wonder certain areas of the world have constant wars.

[Gavin]

We certainly discussed how crap pareto optimality was in my undergraduate degree (graduated 2011) but it wasn't the sort of thing you needed to remember to pass an exam and I doubt most people gave it much thought.

[dBonar]

"optimal", "efficient", "rational" and "irrational" are problematic words. They are often used in specific jargon meanings (and not always the same technical meanings in all cases). Then the results are paraphrased into catchy titles and sound bites where the words have looser, colloquial meanings. When even the professionals are blurring the lines between the technical and colloquial meanings, how can people reading the commentary on the commentary be expected to keep it straight?