Gary takes one more step in the book towards admitting that people are people. He tries to endogenize tastes, instead of just sticking them in the category of things about which we should not dispute. He's still a little dazzled by sociobiology, and inclined to reach for hokey arguments from evolution. And his standard of scholarship in using what he takes to be the findings of anthropology or sociology is not high. Economists who view themselves as theorists, as Gary does, are usually casual about the factuality of their stylized facts. But it's step forward. In his next book I'll bet he finally admits that human beings love each other, too.
What's brilliant about it is what's always been brilliant about Gary. He is, after all, something like the best economist of his generation. About 1974 in the Quad Club at Chicago Ted Schultz, the great agricultural and development economist, a Nobel Prize winner himself from an earlier generation, asked me what I thought was the most exciting trend in modern economics. I said, "The application of economics to things conventionally outside it: Gary Becker."
Ted agreed. We all did. Gary was well-known around Chicago as the best graduate student the place had produced since Friedman and Stigler (I would put Jim Buchanan in that group, but for some reason he was never popular at his alma mater). When Gary was still a student he came up to Al Harberger at a party, Coke in hand (Gary's, not Al's hand: Al's had a scotch), and said out of the blue "You know, Al, children are just like durable goods." And come to think of it, in some ways they are. What an interesting metaphor.
What's the flaw, then? The flaw is that little matter of love. You can call it "solidarity" or "social embeddedness" or whatever. The "S" variable. Economists since Jeremy Bentham have focused on prudence or price, the "P" variable. (Actually of course both these variables are vectors, of things that sociologists like to talk about or things that economists like to talk about.) The focus by economists on "Prudence" has been very, very productive. Call it the Productive variable if you want. Economics can be thought of as the science of Prudence, planning, looking forward. Adam Smith made a lot of it, especially in The Wealth of Nations.
But in his other great book, The Theory of Moral Sentiments, he explained that Prudence alone is not enough. He said in effect,
Look here: people show five virtues, arranged from masculine to feminine, pagan to Christian, which pretty much cover the field. They are Courage, Temperance, Prudence, Justice, and Love. Prudence is practical wisdom, which everyone needs. The others are what Deirdre will 200 years from now call "S" variables. Temperance is internal balance; Justice is external balance. Courage is the characteristically male virtue [though look at the courage of women]. Love is the characteristically female virtue [though look at the love by men]. They are both dangerous, and I'm not going to mess much with them. In the last century [the 17th] people did appalling evil in the name of Courage and Love. We men of the 18th century, a silver age of equipoise, find Courage and Love unnerving. Even commercial courage, Enterprise, makes me nervous. In my book on Prudence called The Wealth of Nations I do not give Enterprise much praise: invest cautiously, I argue, in agriculture. The core virtues of a bourgeois society are the middle three, Temperance, Prudence, and Justice [two S variables and one P]. I gave a course once called Lectures on Jurisprudence which was about justice. The Theory of Moral Sentiments, which was my first and favorite book of two published, was about Temperance, and about the system of the virtues. My point is that my other book, on what you will come to call `economics' is to be viewed within a system of virtues (and, alas, vices). It would be idiotic to try to make Prudence the only characteristic of people. I understand that young fool Bentham is doing precisely that.
As does Gary Becker. And Paul Samuelson. And all our mainstream neoclassical gang of boys playing in their sandbox. (Joined by some analytical Marxist boys recently. Girls of whatever school do not find the sandbox so enthralling.) What's loony about economics and about Gary Becker is specializing in Prudence, against Adam Smith's advice. What's loony is trying to do economics without the S variables.
Smith thought of people as people, not as calculating machines. His people are not very exciting, but they are a step above the monsters of Prudence that modern economics has taken for the only models. Yes, said Smith, people calculate according to Prudence, but within a system of the virtues. A Prudent man who considers Justice, worries about his Temperance, has people and principles he Loves, and yet has the Courage to venture forth is more like a man you might want to live with. (I say this from a woman's perspective. Any tall courageous, temperate, prudent, just, and loving men out there?)
The political scientist Joan Tronto has argued recently that Smith is in an odd way a feminist, in that his system of virtues is nothing like the macho focus on Prudence plus Courage, without an ounce of Temperance, Justice, or Love, that characterizes Monday night football and American business. And it is certainly not the wimpy version in economics that focuses on Prudence alone. Smith's system was based on classical stoicism augmented by one the three theological virtues, Love (as Vivienne Brown argues in her brilliant book, Adam Smith's Discourse). If Smith was made nervous by Love and Courage, he was simply terrified by Faith and Hope ("Faith, hope, and charity, these three; but the greatest is charity," said St. Paul). The awful 17th century could be characterized by the misuse of Faith and Hope out of the three theological virtues. The 19th century got back into the Faith and Hope games, on a simply shocking scale, in nationalism and socialism. We in the 20th century paid the awful price. Smith was no dope. A system of virtues has the virtue of any balanced system.
Taking one virtue out of the system is a big mistake. We social scientists need P and S variables. It's Becker's mistake. Gary is to be congratulated, and to be given Nobel prizes, for showing how far you can go. The self-deconstructing program of rational expectations has the same merit.
You might reply, "Sure, people have other virtues than Prudence. But we can get along on modeling them with Prudence alone. We can get along without the S variables. Leave them to the sociologists."
Sometimes. I am willing to go along with Prudence as the sole guiding principle of finance. An economist like Fischer Black (those who knew him also loved him, and mourn bitterly his recent death) very reasonably applied the most extreme principles of Prudence and invented whole new classes of assets.
But economists don't recognize the limits of Prudence as an all-purpose theory of why people behave as they do. Gary is one leading example. Gary's mentor George Stigler was another. George was fond of claiming that voters voted their pocketbooks. Prudence. Just as Adam Smith said (George was a student of the history of economic thought, but an exceptionally bad one). Departments of Political Science across the land have remodeled themselves into Stigler Studies, which they call rational choice theory. Voters are, say the new political scientists, monsters of Prudence. Don't tell me about S variables. We econowannabes can study voting with P variables alone.
The trouble is, as one can find out by looking at the appendices to George's own, wonderful The Theory of Price (any edition), is that it is of course Imprudent to be at the polls in the first place. Every economist understands this (please explain it to the political scientists, would you?) If Prudence is the only rational behavior, then it is madness to waste any slight inconvenience going to an event over which you have no influence. The probability of affecting the outcome in any moderately large election is nearly zero. What George proposed, and proposed dogmatically with much blustering and sneering, which convinced many timid souls who could not stand up to his bullying, was a theory of Prudence concerning a population of madmen.
The problem is forgetting that people act out of more than Prudence. People go to the polls because they Love their country or are obsessed with Justice. It would be miraculous if one could separate their motives for going from their motives once they close the curtain behind them and face the voting machine. Orthogonality is a happy accident, not something one can blithely specify. The S variables need to be acknowledged along with the P variables.
An old and important case of leaving out the S variables is Thomas Hobbes. I think I have told you before what's wrong with the Hobbes Problem, but let me say it again. Only a gang of boys playing in a sandbox would think it was a neat question to ask over and over again: "Will a bunch of unsocialized brutes form a civil society?" By the Prisoner's Dilemma (a theoretical proposition proven false repeatedly in experiments), no, and the life of man will therefore be solitary, poor, nasty, brutish, and short. Thus Hobbes, and a series of P thinkers down even to the otherwise very sensible James Buchanan.
The trouble is this, as any woman could have told the boys. People are already socialized. After all, they are raised in families. So the correct question is: Will, say, French people form a civil society? You can see that the Prisoner's Dilemma is suddenly less than decisive. Once you let the S variables in, the impact of the P variables changes. Once Love or Justice intervenes, Prudence is not pure. It lives in a system of the virtues, not on its own.
The point can be put econometrically. Suppose you are trying to explain some behavioral variable, B. It doesn't matter what it is: voting, consuming, investing, rent seeking, risk taking, I don't care. Suppose that in the world as we know it the behavior is affected by both P variables and S variables, by both Prudence and by the rest, such as Love or Temperance. The correct specification for regression is: B = Alpha + Beta P + Gamma S + Epsilon.
What happens if you leave out S? Well, unless cov(P,S) = zero, by God's grace, blessed be Her holy name, the pseudo error term Gamma S + E will be correlated with the included variable P. The estimate of Beta will therefore be biased and inconsistent. In cases like Fischer Black's in which the covariance is small, the problem is small. In cases like Gary Becker's--families, voters, criminals--in which the covariance is high, the problem is large. If the degree of prudence of criminals is affected by their unusual lack of Temperance or their weak sense of Justice, then leaving out measures of these will ruin the empirical work. It just will: everyone knows.
Dogmatic P-ers will reply, You can't measure S. Aunt Deirdre replies to them: Who says? Have you tried? And most important, that an important variable is hard to measure is not the end of scientific attempts to do so, at least among real scientists.
I am not saying that studying P variables is a poor idea. In fact the case is entirely symmetric. The sociologists and anthropologists who leave out the P variables and only work on S variables are making the same mistake. Karl Polanyi gave the world in 1944 a program for economics without P variables. No, it's flawed, too. Either diamond needs a setting, or it's just a rare rock.
