Steve Reads

I direct your favorable attention to Is Equality Passé? Homo Reciprocans and the Future of Egalitarian Politics. It’s the first essay I’ve seen that very naturally connects morality, our responsibilities toward those less fortunate than ourselves, and (the prose version of) evolutionary game theory. It’s also an argument about the particular kinds of egalitarianism that we can expect to sell to our fellow-Americans. Well worth the 15 minutes or so that it’ll take to read it.

If you’re interested in that, I would, of course, recommend moving on to Microeconomics: Behavior, Institutions, and Evolution, Bowles’s magisterial textbook. It builds out the arguments from “Is Equality Passé” at a mathematically rigorous level. The book combines substantial empirical backing with ample theory; it’s more than sufficient to rebut the claim that neoclassical economics is the best we’ve got.

A lot of the games mentioned in the books I’ve been reading involve one-shot contests. Hawk-Dove is the most straightforward: Hawks always fight, and Doves always back down. When a Hawk fights a Dove, the Hawk always wins. When a Dove encounters another Dove, you can set up the problem in various ways, but maybe they’re equally likely to take possession of whatever resource they’re after. Likewise for Hawks, though maybe each Hawk suffers some cost to fighting.

Then you can introduce other strategies to Hawk-Dove to model the idea of “ownership.” Maynard Smith uses Hawk-Dove-Bourgeois; a Bourgeois acts like a Hawk if he’s the owner, or like a Dove if he’s an intruder.

You could introduce learning in various ways, with varying degrees of complexity — e.g., in repeated contests, start out playing Hawk or Dove with some fixed probability (let’s say everyone in the population starts with p_0H and p_0D as their probabilities of playing Hawk or Dove), and play Hawk in subsequent rounds with probability proportional to your success playing Hawk in previous rounds. None of the books that I’ve read so far has tried to model anything this complex, perhaps because it would involve using Markov chains before too long; the math would probably get trickier than these introductory books would want.

In most of the models I’ve encountered so far, players understand from the start whether they’re Hawks or Doves — which is why they either always back down or always fight. But a player may not be aware of his role until he’s played (and won or lost) a few rounds. It would be interesting to study whether the always-play-Dove and always-play-Hawk strategies are actually stable equilibria of a learning process.

Or suppose a mutant that has a good read on his opponents enters the population: he can detect whether he will win or lose against his opponents with better accuracy than some large fraction of the existing population, and plays whichever strategy will get him the highest expected payoff from that round. Or weaken that assumption: suppose that the rest of the population cannot detect the likely outcome of any contests, and that the mutant can do so only 1% of the time; now, how quickly will the mutant spread? One presumes that it’ll spread fairly quickly, depending on the payoffs to Hawk and Dove. Now suppose another mutant comes along that can accurately read his opponent’s strength 2% of the time. And so forth.

In practice, we’d expect that if two animals confront one another, and one is bigger than the other, then the smaller one will back down; it’s not worth playing the game when the outcome is certain. If the next best resource for the smaller competitor isn’t much worse than the one he could fight over, then he should take that one. Everything would depend on the cost of fighting, and on the disparity between the contested resource and the next best alternative. But this seems like it might be an interesting model. Given what Bowles taught me, it shouldn’t be that hard to work out.

P.S.: Ah. See the beginning of chapter 9 in Evolution and the Theory of Games:

A number of complexities arise in actual animal contests.

 . . . 

3. A single asymmetry present, but information about it is uncertain (e.g., differences in size or strength). Such contests involve a phase of ‘assessment’. Considerable theoretical difficulties arise if the information acquired during assessment is uncertain.

For a book written by a professional biologist for professional biologists, it’s remarkably clear to someone who just has general mathematical knowledge. It doesn’t hurt that I’ve been reading about similar game-theoretic treatments in economic contexts.

I can’t really keep up with the biological arguments, and there’s enough biological vocab (“pleiotropy”, “linkage disequilibrium”) that some of the details are lost on me, but the general principles stand out. Those principles are straightforward enough: Maynard Smith is explaining a series of tools from evolutionary game theory that are based on simple axioms, and yet unify a wide range of phenomena. He then applies these tools to a set of observed biological phenomena, and shows that they explain the phenomena well. (He’s also scrupulous about pointing out the weaknesses in all the evidence before him, which is a nontrivial part of the book’s charm.)

The biggest such tool is the Evolutionarily Stable Strategy, discussed here earlier in the context of Samuel Bowles’s book. To reiterate: an ESS is a strategy which, if played, cannot be beaten by any invader (if that invader is relatively small; of course a strategy may succeed if the number of invaders dwarfs the existing population). In particular, a strategy is uninvadable if the payoffs to adopting it exceed the payoffs to any of its potential rivals. To me, this carries the same intuitive appeal as natural selection itself.

Maynard Smith’s book patiently, concisely, and almost breezily discusses the evidence for ESSes in nature, including cases like the sex ratio that should interest even non-biologists. The sex-ratio problem is simply: why do there tend to be as many males as females? Maynard Smith assumes that the full range of sex ratios, between 0 and infinity, is available to a species, and tries to find an ESS at which the population would settle. The measure of fitness here — the payoff to the ESS — is how many grandchildren a given mother and father would produce. Assume everyone in the population is “playing” the “strategy” of having as many male children as females, then assume a mutant comes in playing a different strategy. If the payoff to the mutant strategy is greater than the payoff to a 1:1 sex ratio, then the mutant is likely to succeed and displace the 1:1 sex ratio; contrariwise, if 1:1 produces more grandchildren, then it’s an ESS and the population is likely to settle there.

There’s a theoretical tidiness to the whole enterprise, which makes the book a joy to read. Highly recommended — particularly if you’ve not encountered ESSes before. If your biological interest exceeds your economic interest, then I’d recommend Evolution and the Theory of Games even more strongly.

Paul Krugman’s response in an ongoing debate over whether economics quashes anything vaguely out of the mainstream reminds me of something I wanted to mention: Sam Bowles’s book is more an updating of orthodox economics than an overthrowing of it. Nash equilibria, evolutionarily stable strategies (which I guess isn’t really part of orthodox economics, but there’s no reason why it would have trouble fitting there), Pareto optima and all the rest still play a central role. People are still assumed to be maximizing something or other. It’s just that when they do maximize, they sometimes maximize for non-rational reasons (e.g., revenge — though again, Robert Frank gives us every reason to believe that revenge is rational). Sometimes their utility functions reflect their coworkers’ or families’ utility functions. Bowles wants to keep what works in economics, but use a more accurate psychology.

P.S. (21 July 2007): Bowles’s longtime collaborator Herbert Gintis reviews a critique of neoclassical economics; his remarks are relevant here. I include them below.

(more…)

What an extraordinary read. The shortest way of summarizing Bowles’s book is that it was a synthesis, and a deepening, of every behavioral-economics work I’ve read before now, with substantially more theoretical support; it supersedes all those earlier works. It embeds all of them in a rich economic structure that nearly takes institutions as a given and works within that structure; it only avoids completely taking institutions as given because it knows that so many of its readers are used to an economic framework that ignores them.

I’ve already said most of what I want to say about this book. I’ll just reiterate that it should be required reading for anyone steeped in the standard economic orthodoxy. In particular, Bowles should write a version of this book aimed at Econ 101 students, to keep their minds more open than those classes tend to. College students with an introductory econ education are frightfully likely to become Objectivists; avoiding this outcome would start us down the road to a social optimum.

Bowles’s Microeconomics has hit me like a revelation. It deeply and grandly reconnects economics with political science and with moral theory, and gives a sound mathematical and empirical basis to both. It steadfastly resists dogma, even while arguing persuasively for behavioral economics and a realistic socialism. It is a work that could only have come from someone who’s studied the issues and listened patiently to his opponents for many years.

Bowles’s fundamental contention is that it doesn’t make sense to study economic interactions in isolation; such an interaction cannot be understood outside of the institutions that make it possible. The simplest example here would be the free market: markets exist because the government will defend your private property with guns; because the government will enforce contracts; because private property is freely exchangeable by such contracts; etc. Without those protections, there would be no free market.

So Bowles zooms in. Consider economic actors behaving inside of the market. They are participating in a game that evolves over time by rules that the institution sets. But at the same time, an institution can be viewed as the endpoint — equilibrium — of a game; institutions themselves change over time, for at least a few reasons that would be valuable to study:

1. Invaders from outside the institution. Consider, for instance, a nomadic bartering society which has just been attacked by a Western capitalist power, and consider what effect the invasion will have on the bartering society’s existing institutions. Will barter survive, or will it be supplanted by the market? (Of course we’d need to consider that the invader may have more guns. This may or may not be a part of the economic model itself.) Under what conditions could barter be expected to continue?
2. Technological change within the existing institution. Consider the effect of the printing press within a pre-industrial society.

Bowles makes the case strongly — if implicitly — that ignoring institutions, or more specifically treating them as a settled affair, has left the most interesting problems beyond the reach of economics. Why do institutions change at all? Under what conditions will an institution resist invasion by novel institutions? By asking questions like these, we can turn economics into the quantitative study of problems that interest all humans — not just the very limited problem of how a market with alienable property and complete contracting will evolve.

To address these problems, Bowles has a number of tools at his disposal, but two important ones are worth calling out:

1. Evolutionary game theory, particularly the Evolutionarily Stable Strategy. An ESS is a strategy that will successfully fight off invading strategies. ‘Belief in a supernatural being,’ for instance, may be a strategy, with atheism viewed as an ‘invader’. (Without thinking too hard about it, it seems that we’d have to explain the empirically observable fact that theism is an ESS. I don’t quite know whether a strategy is an ESS if it will survive as time goes to infinity, or whether finite time is what we need.)

   Bowles gets tremendous power out of the simplest models. Imagine we’re divvying up some shared resource, like a large animal that we’ve all just felled in a group hunt. Suppose there are three kinds of people in our society: Grabbers, Sharers, and Punishers. Sharers will try to share any resource before us. Grabbers will always try to steal it, and we can state some conditions under which they’ll tend to dominate Sharers. Punishers, when faced with a Grabber, will team up and try to fend him off, with a probability of success that increases with the number of Punishers in the group; when there are no Grabbers around, Punishers act like Sharers. If they succeed in Punishing the Grabber, they will split his stolen goods evenly amongst themselves.

   Now, with those stipulations, under what conditions will the society fill with Punishers? Under what conditions will Grabbers come to dominate? The answer clearly depends on how many of each type we start with, though it may be that Punishers will always dominate Grabbers, and that over time the proportion of Punishers in the population will rise. The initial intuition is that Punishing dominates Sharing if there are enough Punishers in the population, because Punishers get just as much as Sharers do when faced with Sharers, and also have the potential to get resources from Grabbers. The answer, though, depends on the exact numbers.

   What does any of this have to do with anything? The great beauty of it is that in just a few pages Bowles has sketched out a quantitative reply to the debate between Hobbes and Rousseau. Under appropriate conditions, a culture based on mutual punishment for theft can, in fact, persist and remain stable against invading strategies — a reply to Hobbes. Of course any realistic model would have to factor in far more variables than this, and would have to consider far more types of people than the three we’ve listed here. But it’s a start. And it’s empirical, which is an improvement over mere thought (‘The method of “postulating” what we want has many advantages; they are the same as the advantages of theft over honest toil’). And it extends economics far beyond the interaction of homogeneous actors in financial markets.

2. Incomplete contracting. It has been known since at least Herbert Simon’s paper “A Formal Theory of the Employment Relationship” that employment agreements cannot be treated like contracts for the exchange of commodities: in the latter case, the contract can completely specify what I will get out of the trade, and what I will give you in exchange. Employers, by contrast, cannot perfectly monitor your or my work; the contract is incomplete in that sense. Incomplete contracting, in Bowles’s analysis, gives rise to a great range of institutions; all are based in some way or another on trust, and how we respond when trust is missing. The most straightforward modification of the complete-contracting model to employment is the contingent-renewal model, in which I incompletely contract with you for a job, I hold that job for a while, and at the end of a period of time you renew the contract if you think I’m doing a good job — or fire me if you don’t. This seems to be a much more realistic model of how employment works, and Bowles uses it to good effect.

   I should mention, in the employment-contract context, that Bowles is gifted at telling me things that I should have thought of on my own. E.g., if employment could be completely contracted, then why would a boss be necessary at all? And he praises the Fundamental Theorem of Welfare Economics — namely that under certain precise conditions, the free market implements an economic optimum — but thoroughly refutes its application beyond those conditions. For instance, the Theorem requires that economic actors be price-takers (i.e., they exist in a large, undifferentiated market, and do not have the power to set prices). He then notes the standard question that every Econ 101 student should have (but I’m not sure that I did have): if no one has the power to set prices, then how do prices change?

Bowles’s book is invaluable for

1. Studying the growth and change of markets, where Econ 101 would tend to depict markets as settled problems with static equilibria;
2. Introducing the reader to a host of tools that are valuable well beyond economics, particularly those of evolutionary game theory;
3. Extending the scope of economics to moral and political theory, where its roots lay.

It is an absolute masterpiece, which I would recommend to anyone who prefers to think concretely.

I’d like to request that we stop using a particular phrase. Apparently Mitt Romney “bob[bed] and weave[d]” on whether he “believes in evolution.” Let’s not talk about it that way. Let’s ask people whether they’ve taken enough of a look at the evidence to decide for themselves whether they believe that evolution by natural selection is a good hypothesis. By that standard, I don’t actually think that most of us have any right to believe in evolution. We don’t believe that the world was created by a single act of God’s will, but we’ve also not questioned the science enough.

Now, we’ve not questioned a lot of what we say we know, because we rely on people we trust to tell us what the world is actually like. This is the way it should be: the world is a complicated place, and we can’t all be expected to go hang out on the Galápagos Islands studying evolution in action. So if we actually wanted to be honest about what we’re asking Romney, we’d ask him, “The people you trust to tell you the way the world is  . . .  have they led you to believe in evolution by natural selection, or not?”

But let’s be clear: by any reasonable standard of evidence, evolution is not just a theory; it is an actual, observable fact, as well-established as anything else we can see with our eyes. It is not something that only happens over millions of years; it is something that happens on very short time scales, when animals with slight anatomical differences survive and reproduce at different rates. It’s not even sensible anymore to ask someone whether he “believes in” evolution.

Doing so, in any case, cheapens the science. Science shouldn’t be something you just “believe in.” What they want to ask candidates is: “Has the theory overcome your hopefully high level of skepticism and attained provisional truth?” (There’s another level where it does make sense to talk about “believing in” science — namely, at the very deep level of our beliefs that’s largely beyond change. For instance, I suspect there’s very little anyone could do to make me a devout Muslim.)

Of course it’s a political problem; what they’re asking is actually the reverse of what they mean. What they mean is: do you subscribe to a particular brand of American fundamentalist Christianity that rejects much of modern society, including the theory of evolution? Evolution becomes a shibboleth: it’s a quick test of your loyalty to a particular set of ideas. It’s a totem: it’s probably not very important as a belief on its own, but rather important for the set of beliefs that it points to.

So then, finally, all they mean to ask him is: do you believe in the teachings of fundamentalist Christianity, or don’t you? The American political climate forces a binary choice here, and that binary choice becomes self-sustaining: maybe you’re somewhere between fundamentalist Christianity and a belief in natural selection, but you don’t want to be associated with those people, so you move to the other extreme. More and more of your compatriots do the same, and eventually believing in the other side’s views is something that only an isolated minority within your group does [*], with predictable consequences for your social life and sex appeal (“Oh, you don’t believe in evolution?”). For a similar example, try hanging out with a group of liberals and explaining that you have some reservations about abortion, or about partial-birth abortion in particular. Try expressing any reservations at all. You will not get far. (I’m not picking on liberals; flip that sentence around in the expected way for a group of conservatives.) Partly — mostly? — it’s because your group believes that if you cede even an inch of ground to the other side, they will take a mile. The mechanisms of the slippery slope are at work here, among others.

Certain issues become so polarizing that pretty soon your group defines itself by its stance on them. I don’t deny the wisdom of this practice, actually: belief in evolution or in a woman’s right to have an abortion probably count as “signals” in the economic sense. Because it’s hard to get information on what someone is “really” like, we rely on measures of their personality that we can see — particularly those measures that are difficult to fake. (I only realized the difficult-to-fake criterion by reading Passions within Reason.) In this society, there’s a fairly high cost — namely social ostracism — to asserting specific political beliefs within specific groups, so your willingness to admit that you oppose abortion would probably count as a costly-to-fake belief; you wouldn’t say it unless you really meant it and were willing to stomach the consequences.

This naturally brings up the question of why certain issues are so polarizing that they can ever become signals of a group’s beliefs. American politics has a few issues that are this way (abortion, gun control, evolution), and many that aren’t (universal health care, seat-belt laws). Certain issues seem to have the capacity to become totems, while others don’t. Understanding what makes these issues polarizing would be highly illuminating. Why is abortion a bigger deal in the U.S. than it is in Europe, for instance? I have my hypotheses, but they’d be out of place here.

It does seem fairly certain that when an issue becomes as polarizing as these, the capacity for productive policy-making has disappeared. It is no longer possible to have a civilized discussion over abortion; you’d either prefer that rape victims get back-alley abortions, or you enjoy committing third-trimester babies to death. When groups can only talk past one another, it’s time for a change.

[*] — with more than half of Bowles’s Microeconomics under my belt now, I should build a time-evolving model of this process. It wouldn’t be that hard, in all likelihood.

Because this is an all-Cosma-recommendations-all-the-time blog, I should mention that I started reading the above-named microeconomics textbook last night, on Cosma’s recommendation. (To do so, I had to put down John Roemer’s A Future For Socialism for a moment. Note that Cosma’s review of AFFS trumps both its Harvard University Press homepage and Amazon’s page for it. I yearn for the day when I have that kind of Google cred.)

I have almost nothing to say about it just yet, given that I’m only through the prologue. It looks like it will be a good synthesis of the last 20 years of economic research, moving beyond the models of purely self-interested economic man and into something that accords better with experimental evidence.

It also looks deliciously multidisciplinary; the bibliography is quite hefty. I’ve had a post in my head for a while, wondering whether specialization — the academic sort in particular, but industrial and so forth as well — is likely to be counterproductive. On the one hand, it allows people to become domain experts, but on the other hand people are likely to ignore research from neighboring fields that might help them. And each discipline develops a jargon that quickly becomes impenetrable to outsiders, which makes it even harder for others to draw on its research. I hope Bowles’s book lives up to its strong multidisciplinary promise.

But actually, the whole reason I started writing was for a much more cosmetic reason: Bowles’s book is gorgeous. The cover is Ambrogio Lorenzetti’s “Allegory of Good Government In The City,” and all the text — including the titles on the cover, I believe — is set in Sabon. It’s a joy just to read it. I hope and expect that the text itself lives up to the design.

(Yes, I’m a book-design fetishist. So sue me. I read a lot of books.)

Incidentally, there’s another “Allegory of Good Government” that appears on the cover of the Penguin edition of Ruskin’s Unto This Last And Other Writings. The Wikipedia explains.