Book Review of
The Myth of the Rational Voter: Why Democracies Choose Bad Policies
Princeton University Press, 2007, ISBN-13: 978-0-691-12942-6
Peter A. Taylor
August 9th, 2007
Of all the ways of defining man, the worst is the one which makes him out
to be a rational animal.
— Anatole France
"Stories without fools"
Bryan Caplan is disappointed with democracy. He gives trade protectionism as an example. Economists have been arguing for 200 years that trade protectionism is bad for the economy, voters have said for 200 years that they want a good economy, and for 200 years democracies have had trade protectionism. In relative terms, this isn't so bad. It beats having things like the Berlin Wall. But the reason East Germany built the Berlin Wall is obvious. Why do we have trade protectionism? Caplan writes that trade protectionism is "less appalling than the Berlin Wall, yet it is more baffling."
There are a number of ways to explain this. One answer is that maybe democracy is working better than we give it credit for: maybe economists are wrong and trade protectionism is good. Taking the textbook arguments as read, Caplan spills a great deal of ink in chapters 3 and 8 arguing that what economists believe is typical of educated people in general, and they are not plutocratic apologists or members of some de facto religious cult. A second answer is that the government doesn't do what the voters want, either through inefficiency or corruption. This answer doesn't work either, at least in the case of trade protectionism. Voters like trade protectionism. A third answer, which economists generally defend, is that there is something wrong with the voters. But these are the same people who participate in the economic markets that economists defend, and whom economists routinely and almost automatically assume are "rational." In Robin Hanson's terms, economists tell "stories without fools." Textbook arguments about why economists have gotten away with assuming this in the past are beside the point. If we're interested in practical political questions such as whether we should increase or decrease the authority that a regulatory agency has over an industry, it's important that we are consistent in the assumptions we make in analyzing the behavior of markets versus governments. This is a major part of the motivation for economists to study political institutions in the first place, the difficulty of comparing markets, analyzed by assuming people are selfish, against institutions analyzed by assuming people are civic-minded (besides the suspicion that the political scientists' orchard still has some low-hanging fruit). If we want to be intellectually honest, we have to be consistent in our assumptions.
Stories with fools
The usual way that "public choice" economists (the ones who study political institutions) explain the perverse behavior of voters is in terms of "rational ignorance." The idea is that people use the information they have well, both as voters and as consumers, but they have less information about political issues, and less incentive to seek out more information, than they do about their regular day-to-day affairs. But is ignorance a sufficient explanation? Caplan's answer (p. 109) is no: "Rational ignorance, long since convicted by a vast literature of subverting democracy, lacks the means to commit the crime of which it stands accused. ...[I]gnorance cannot carry the weight the critics of democracy assign to it." Caplan blames much government misbehavior not so much on voters' ignorance as on voters' cognitive biases, which are expressed in an environment in which individual voters have no appreciable incentive to engage in the mental discipline needed to correct them. He calls this "rational irrationality."
The short version of his thesis appears on p. 3 (quoted here):
To go any further into this controversy, we have to examine the different incentives people face as voters versus shoppers, both in terms of ignorance and mental discipline. What do people want? What strategies are open to them? How effective are the different strategies? The most convenient assumption economists could make is that people are interested in tangible things that they can buy with money and enjoy independently of other people. In politics, Caplan calls this the Self-Interested Voter Hypothesis (SIVH). But as Caplan argues, people also like to flatter themselves. People also come with "default beliefs" that require effort to overturn even if they are not particularly cherished, and people generally want to avoid going to unnecessary effort. Some of the strategies that Caplan is concerned with involve whether or not to invest time and effort, and risk embarrassment, either acquiring information or engaging in the mental discipline needed to make a technically correct decision. But he is also interested in the degree to which people make decisions based on their egos rather than based on their wallets. So far the analogy between voters and shoppers holds up. My ego might get in the way of my buying a foreign car just as it might get in the way of supporting NAFTA. But when we address the effectiveness of these strategies, we have a problem Caplan calls "decisiveness:" what is the probability that a single decision-maker's decision will determine the outcome that he receives? As a shopper, my decision to buy a Ford instead of a Toyota has a 100% chance of resulting in me buying the Ford. As a voter, the chance that a major election will come down to my one vote is astronomically small. For practical purposes, it is zero.
One immediate consequence of this logic is that the Self-Interested Voter Hypothesis makes no sense. The result that I get from voting is the effect that my candidate winning would have on my wallet, times the negligible probability that the election will come down to one vote, plus the effect that the act of voting has on my ego. I thus have no effective way to vote my wallet. All I can do is vote my ego (ie. "conscience")1. Caplan presents empirical data that confirm this (for example, contrary to popular belief, men are statistically more pro-choice on abortion than women are).
The same logic works whether we are talking about ignorance or mental self-indulgence: the lack of effectiveness of a single vote in deciding an election means that, as voters, we have no material incentive either to be politically well informed or to exercise any mental discipline. Ordinary voters may read and think about politics for entertainment, or to impress people at cocktail parties, but not for financial gain.
Fallacies of Composition and Division
Many people, exposed to this line of reasoning, respond with a fallacy of composition (or division), confusing the properties of a set with the properties of the members of the set. A textbook example of one of these fallacies would be: G. W. Bush is a Republican, Republicans are numerous, therefore G. W. Bush is numerous. In the case of voters' incentives, the fallacy runs: The nation is collectively better off if people vote rationally, all of the voters are similarly situated, therefore each voter will be better off if he votes rationally. Something in the human brain expects there to be a law of conservation of incentives, but there isn't. Instead, we have the Tragedy of the Commons. The political arena is not a market, it is a commons. My ego is affected by my behavior. My wallet is affected by the behavior of the majority of my neighbors. We all suffer because none of us has a material incentive to behave responsibly.
But this reasoning works equally well for rational ignorance or for "rational irrationality." Both voters' ignorance and human irrationality are well documented. How does Caplan know that the problem is irrationality rather than rational ignorance? Caplan has two main arguments, one involving what he means by "rational," and one involving gullibility.
Rationality: I am trying to avoid a long discussion of semantics, but part of what Caplan means by "rational" is that any errors are random and unbiased. Economic theoreticians love to assume this, because if people's errors are unbiased, they tend to cancel, and are easier to ignore. This is sometimes called "The Miracle of Aggregation." Caplan gives an example of a contest in which 787 people guessed at the weight of an ox, and the average was off by 1 lb. Similarly, if voters are equally likely to guess high or low at the optimum amount of government regulation, and if elections are carried by the candidate supported by the median voter, then the voters collectively should be able to get things pretty much right. 99% of voters may be completely clueless (which Caplan argues is not far from the truth), but if their errors cancel, the cluefull 1% in the middle will call the shots. But if errors are systematically biased to one side, such a small number of cluefull voters is nearly irrelevant, and the median voter's opinion will closely reflect the average bias. Caplan documents four important kinds of bias in economic opinions, which he calls anti-market bias, anti-foreign bias, make-work bias, and pessimistic bias2. David D. Friedman has an interesting paper on evolutionary psychology which goes nicely with this book, in which he argues that people are predisposed to expect "thin" markets, where prices are largely set by bilateral monopolies rather than by supply and demand, predisposed to discount future rewards in a way that defies economics textbooks, and predisposed to value their current possessions excessively. To paraphrase Edward O. Wilson: we evolved to believe in monopolies; we didn't evolve to believe in the law of supply and demand. See also Will Wilkinson.
Gullibility: There are a number of workarounds for dealing with one's ignorance, but a rationally ignorant voter's principal recourse is skepticism. Rational voters should be no more gullible than rational used car buyers. Used car buyers know that the seller knows more about the car he is selling, and why it was put up for sale, than the buyer does. Knowing that the game is rigged against them, rational buyers bid conservatively, and used cars change hands less frequently than they would in an ideal market. Similarly, if I am thinking of having my house remodelled, but believe that the job is worth doing only if it is done well, believe that poorly supervised contractors are unlikely to do what I want, and know that I am too busy to supervise contractors closely, then it simply doesn't make sense for me to remodel. There are many jobs that are worth having the government do only if they are done well. A rational voter would realize that they are not going to be well supervised, and hence are not likely to be done well. Rational voters would vote against any proposal for government to perform a task unless it is worth doing even if it is done badly. But this is not how real voters behave.
Caplan offers some minor arguments as well, including introspection, the personal testimony of people like Whittaker Chambers, and the nature of the errors. He writes (p. 99): "The sillier errors get, the more likely it is that the cause is lack of mental discipline, not lack of information."
Not all of the implications of Caplan's reasoning are obvious. The demise of the Self-Interested Voter Hypothesis (SIVH) indicates that voters should be more altruistic3 than shoppers. This sounds like a good thing, but is it? My response is that this kind of "altruism" isn't genuine concern for one's fellow man, but merely pretentious intellectual self-indulgence4. Caplan's response is that altruism allows a group of people to act with less internal friction, but does nothing to make them more sensible. Altruism thus makes the good better and the bad worse. Altruism combined with irrationality is arguably like grease on a guillotine.
In addition, the demise of the SIVH invalidates the usual arguments for attempts to increase voter turnout. Better educated people, who are less prey to the cognitive biases Caplan identifies, are over-represented among regular voters. Increasing voter turnout dilutes the benefits of this.
From a theoretician's standpoint, the demise of the SIVH makes much of classical public choice theory redundant. False beliefs simply produce bad policy. There is "no need to build a rickety bridge from the public interest to each individual's private interests."
Caplan writes (p. 124), "One interesting prediction of rational irrationality is that fluctuating incentives make people bounce between contradictory viewpoints." For example, people in poor countries will try to emigrate to rich countries, but will vote against adopting the policies that make countries rich. The private choice is decisive. The public choice is not.
The tendency of voters to have "mixed preferences," such as simultaneously wanting trade protectionism and a strong economy, puts politicians in a much more sympathetic light than they usually enjoy. None of your options as a politician are attractive. Voters don't like being lectured to. Should you defy them by doing the right thing? Screw your successor? Blame the other party in a divided government? Blame subordinates5? Honesty doesn't pay. But people can often "detect dishonesty from body language, tone of voice, and more." So the more successful politicians tend to share the public's irrational beliefs6.
Politicians are demagogues, pander, and break promises. To a great extent, they have to, because of the way irrational voters vote. On the other hand, politicians do have some "slack" or "wiggle room." This is both good and bad, and it is on how they use this slack that we should judge them. Caplan writes, "When a master does not know his own interests, a disobedient servant can be a blessing." It would be worse if the general public consistently got exactly what it asked for. They are often protected by "felicitous hypocrisy." Caplan writes regarding suppliers of political leadership, "Supply side chicanery is only unambiguously harmful given conditions--full voter rationality--under which it does not arise."
According to Caplan, similar remarks regarding "slack" apply to the news media. Like other public choice economists, he regards the news media first and foremost as part of the entertainment industry. (Voters get information about the government, which is a public good, as a side effect of being entertained, which is a private good.) The media mainly tell their audience what the audience wants to hear. There are narrow limits to what members of the media can produce without losing their audience. Caplan notes that the widespread economic misconceptions which interest him predate the existence of the mass media. In any case, rational voters would be too skeptical for the media to be able to do the harm of which their casual critics accuse them.
[Update, 12-7-2009: See Appendix A for a more thoughtful discussion of the news media.]
Well-funded lobbyists also have limited influence over events. The amount of money spent on political advertising is small compared to GDP and only weakly effective. Special interests depend on voter irrationality and need to confine themselves to low-profile issues or low-profile aspects of high-profile issues (ie. apply trade protectionism to this industry rather than that one). Rational voters would interpret a lavish political advertising campaign as a sign of corruption. Caplan writes of lobbyists working along the voters' "margins of indifference."
What to do?
Caplan is not optimistic about being able to fix the problems with democracy. He mentions the possibilities of trying to improve voters' education or trying to bias elections further toward the more educated voters, but notes a catch-22: "Once you use up your political slack, the only way to curtail the political influence of the economically illiterate is to convince them it is a good idea." Which is the problem in the first place. Caplan's advice, then, is for economists to do a better job using the slack they have. The good news, he argues, is that economists, as teachers, have a considerable amount of slack.
The bad news is that economists face a hostile environment, and that this hostility is quasi religious. He establishes the principle (p. 2) that "we turn off our rational faculties on subjects where we don't care about the truth," but he quickly turns to religion as an illustration, quoting Nietzsche (p. 15), "'Faith' means not wanting to know what is true." He writes, "Once you admit that preferences over beliefs are relevant in religion, it is hard to compartmentalize the insight," following this with quotations from Gustave Le Bon, Eric Hoffer, and George Orwell on the essential one-ness of religious and political irrationality. Caplan writes (p. 16), "Political/economic ideology is the religion of modernity." He concludes this section (p. 16): "Natural scientists have long known that the majority disbelieves some of their findings because they contradict religion. Social scientists need to learn that the majority disbelieves some of their findings because they contradict quasi religion."
Religion comes up again in chapter 8, in which Caplan defends economists against charges of "market fundamentalism." He argues that this is almost entirely a case of projection on the part of adherents of the "religion of democracy" (whom I will call "populists"). I say "almost" entirely projection because there are a few individuals, such as the late Murray Rothbard, who may be fairly accused of turning a blind eye towards the defects of markets. But these people are academic circus freaks. Yet populists like Robert Kuttner (Caplan gives numerous other examples) blame the world's problems on a vast and powerful cabal of supposed market fundamentalists. William Greider and Ian Shapiro, among others, are quoted respectively making statements about "childhood faith" in democracy and circular philosophical arguments about the immunity of democracy from criticism. The archetypal populist statement is a reported Al Smith quotation, "All the ills of democracy can be cured by more democracy." These people are respectable and numerous. Respectable economists wouldn't be caught dead making analogous statements about markets.
How does one teach in the face of such quasi religious prejudice? One answer is to emphasize areas of agreement among economists rather than encouraging students in thinking that nothing in economics is settled. Another piece of advice is to emphasize the basics and not to be afraid of oversimplification. Caplan proposes "the Laffer curve of learning: They retain less if you try to teach them more." Finally, he says to make economics "cool" by using better rhetoric: portray popular misconceptions as "the emperor" and encourage students to take the role of the boy who speaks up about his nakedness. In other words, teaching is less about transmitting information than about salving students' egos.
Caplan also has advice for people writing economics papers. Irrationality can be incorporated into economics, but requires more empirical data. Not every model can be a "story without fools." In politics especially, it is essential to take irrationality into account. For example, when considering how much authority a regulatory agency should have, one needs to consider the voters' cognitive biases in guessing at how that authority is likely to be used. An insurance market that could benefit from regulation by a wise, benevolent dictator is likely to suffer from runaway "adverse selection" if voters confuse insurance with charity and decide that it is unfair to charge high risk customers higher premiums than low risk customers. For another example, Caplan praises a paper by Kuran and Sunstein that tries to explain public panics such as Love Canal in terms of media pursuit of ratings combined with a human cognitive tendency to overestimate the probability of memorable events. He also points out that economists are not the only professionals that struggle with voters' cognitive biases. For example, voters tend to say, "A poison is a poison," where toxicologists tend to say, "The poison is the dose."
Caplan's advice is partly a dope slap aimed at other economists, but it
also suggests that political scientists (at least one I talked to a few
years ago) have been right all along in saying that public choice economics
is not ready for prime time. He concludes, "A proverb tells us that 'a wise
man learns more from a fool than a fool learns from a wise man.' By closing
their eyes to fools and folly, the wise men of social science have
artificially hobbled the advance of their own learning."
1. I think of voters' incentives as having the following form:
Payoff = 0.000001 * self interest + 0.5 * bragging rights + 1.0 * moral validation .
0.000001 is a rough guess at the probability of a major election coming down to one vote (and that a particular voter's one vote produces a different result from what a tiebreaker would have produced).
By "self interest," I mean tangible returns that a voter would not have gotten otherwise.
0.5 is a guess at the odds of a voter's preferred party winning the election.
A voter gets "bragging rights" if his preferred party wins. Some voters seem to derive pleasure from being on the winning side, regardless of what the issues are.
"Moral validation" (or "conscience") is the emotional gratification a voter gets from voting for his preferred side, regardless of who wins.
One possible objection to this line of reasoning is that many people's morality may be hard to distinguish from their self-interest. I suspect that the moral validation a voter gets from voting comes largely from a politician telling him that he deserves the tangible returns the politician is promising him. For example, a politician promises me bread and circuses. I vote for him not because I think my vote will determine whether or not I receive bread and circuses, but because I enjoy the thought that I deserve bread and circuses. But the result is the same.
So in many cases, SIVH may actually be a fairly decent proxy for Caplan's "rational irrationality." But this is only true if the tangible returns are obvious and not subject to cognitive biases. A minimal amount of complexity or deception should be sufficient to get voters to vote against their interests.
2. Caplan's major categories of economic cognitive biases are: anti-market bias, anti-foreign bias, make-work bias, and pessimistic bias. Caplan doesn't talk about evolutionary psychology, so those comments are mine.
Anti-market: People systematically underestimate the advantages of markets. People tend to judge one another by their apparent intentions rather than by the consequences of their actions. Selfish=bad. QED. Adam Smith's "invisible hand" is profoundly counter-intuitive to most people (or Caplan's "rickety bridge," if you prefer). Tradable permits are especially counter-intuitive. Part of the problem is that people tend to view profit as a pure transfer, like charity going from the poor to the rich. F. A. Hayek taught that prices and the associated profits served three essential and hard-to-duplicate functions: they provide people with incentives to produce, the capability to produce, and information about relative scarcity. These may be of little importance if you're a nomadic hunter-gatherer to whom wealth is something that literally grows on trees, but they are critical in a modern economy. But many people just see profits as transfer payments. Interest payments (usury) is a special case of seeing profits as transfer payments to the rich. Prohibitions against usury are very common and very destructive. Lay people also tend to favor monopoly theories of price rather than the law of supply and demand, often causing voters to support government interventions that create real problems as a side effect of trying to solve imaginary ones.
Anti-foreign bias: This is partly xenophobia. It is also partly the result of the confusion of money with value. It is also partly the result of viewing trade as a zero-sum game. It is partly not understanding comparative advantage. People tend to think of trade with neighbors in fundamentally different ways from how they think of trade with foreigners (perhaps respectively as expressions of altruism vs. hostility?).
Make-work bias: People tend to underestimate the benefits of conserving labor. They tend to measure productivity by the number of hours worked rather than by the amount produced. There is enormous hostility to downsizing. The Luddites are alive and well.
Pessimistic bias: People tend to look back to non-existent "good old days." Partly, this is because overall improvements tend to be gradual, but there is always some local problem cropping up, to which people pay disproportionate attention.
For a more thorough discussion of these biases, see this excerpt from Caplan's book in the October 2007 issue of Reason magazine.
See also this essay by Arnold Kling on
People Hate Economics. People should think about economics using the
part of the brain they use to play billiards, but instead, they think about it
using the part they use to get a prom date.
3. Here's my idea of a good operational definition of
A large part of altruism, even when it is perfectly honest, is grounded upon the fact that it is uncomfortable to have unhappy people about one. This is especially true in family life. A man makes sacrifices to his wife's desires, not because he greatly enjoys giving up what he wants himself, but because he would enjoy it even less to see her cutting a sour face across the dinner table.
4. As Jamie Whyte put it,
Modern politics is just as you should expect it to be when votes are cast by ignorant people taking advantage of a low-cost source of emotional gratification.
5. The British comedy series, "Yes, Minister", portrayed a role reversal in which the members of the permanent civil service are often the actual political decision makers, with the elected officials looking on helplessly. Mencius Moldbug took this same view in some of his writings, specifically referring to "Yes, Minister" in the context of Ron Paul (the Plum Book) and Sarah Palin.
Caplan rejects this view, claiming that these subordinates really are
subordinate, but that they have been hired by the politicians in order to
serve as cutouts. They are to some degree professional scapegoats,
purveyors of plausible deniability.
6. Tyler Cowan wrote in "Self-Deception as the Root of Political Failure" (2003),
Parties will attempt to exploit voter self-deception and indeed we can think of parties as organized vehicles for this purpose (among other purposes). In particular, parties will try to buy off extremists by making it easier for them to self-deceive.* The easier the extremists can self-deceive about being happy with a party, the more easily that same party can move to the center and capture enough votes to win the election. Towards this end parties will look for symbols that have meaning to extremist groups but do not offend moderates.*Glenn Reynolds calls this "fanservice." — PAT
Appendix A (12-7-2009):
On reflection, and after receiving quite a bit of criticism from a church group to whom I presented this material, I have concluded that Caplan's view of the media is naïve. The American mainstream news media are far more of a monoculture than I can explain in terms of large numbers of independent businessmen pursuing profit in an intellectually diverse society. This monoculture is also clearly to one side (left) of the median voter. Furthermore, even if you do view the media as part of the entertainment industry, all that does is raise questions about Hollywood's reputation for making leftist propaganda films that predictably lose money. Especially given that voters are not rational, it is worth taking a closer look at the media (Ed Driscoll).
1. One explanation of media monoculture is that it is a monopoly or cartel. The textbook definition of a natural monopoly is that average production costs fall as a firm's production increases until one firm produces the entire output of the industry. (A natural cartel has slightly less pronounced economies of scale.) The cost of printing the second and subsequent copies of a newspaper is vanishingly small compared to the total cost of printing the first one. Similarly, two newspapers can halve their considerable field reporting costs if they can share the output of one reporter. This sets the stage for a cartel. For many years there were only three major television networks in the US. The wire services (AP, Reuters) are also candidate villains here. So is the fact that newspapers in several cities can be owned by the same company (NYT, Boston Globe). In any case, copying is cheaper and easier than producing original work. This is true whether you are copying from the NYT or from a politician's or lobbyist's press handout. So it is possible that the news industry is much more monopolistic than it appears to be.
Update, 1-2-2013: Michael Jennings observes that print media have traditionally been dependent on classified advertising revenue that is subject to "network effects". Sellers want to advertise where disproportionate numbers of buyers are, and buyers want to read where disproportionate numbers of sellers advertise. This tends to make newspapers natural monopolies in any given city.
2. A second possible explanation is government coersion or legal influence. Mencius Moldbug writes:
Twice in the 20th century, the US press was gleichgeschaltet as a government agency: once as the CPI, once as the OWI. Significantly, this was done not by appointing some general to tell journalists what to say, Göbbels-style, but by bringing journalists themselves into government. The legacy of these coordination events is more or less what we mean when we say "mainstream media."
Once radio and TV broadcasting became important news media, regulation by the FCC came into play. Since broadcast frequencies are not legally recognized as property, which is protected by the 5th Amendment takings clause, but instead depend on government licenses, any broadcaster who attracts too much negative attention from the government is in danger of being shut down without compensation. The safest policy is to (A) pretend to be professional and objective, and (B) not offend the powers that be in the government. There is a certain inherent dishonesty in pretending to be objective in making decisons about what to cover, what information to include, and how to slant it. In practice, it means only broadcasting material that is similar to the competition.
The "Fairness Doctrine" made it especially hard for news broadcasters to be honest about their biases or to depart from opinions that were popular with the government. This doctrine was discarded in 1987, resulting in a split between the pre-1987 established "mainstream media" and much of the post-1987 media such as AM talk radio.
There is also the matter of access. CNN admitted to biasing its coverage of the Second Gulf War in order to protect its access to Saddam Hussain's Iraq. The Obama administration has tried to punish FOX for unfavorable coverage by declaring it not a legitimate news organization and denying it interviews.
3. A third possible explanation is that certain news outlets (NYT, Washington Post) maintain their dominant position as agenda setters illegally, in essence by selling state secrets in a symbiotic relationship with some of the less scrupulous Washington insiders. The most obvious example is Mark Felt as "Deep Throat" in the Watergate scandal. In exchange for "leaking" classified information and other government secrets, the insiders get favorable news bias and de facto control over the news media agenda. Mencius Moldbug writes,
Effectively, the business of the Times as an institution of journalism can be described as the business of stealing secrets from the public trust, and selling them.
Update: Moldbug also has a nice discussion of the Pentagon Papers here.
In government, selective disclosure creates a power network between the press and its sources. This network does not produce money, but just power. The power is shared between the sources and the journalists. The whole system is about as transparent as mud.
4. The most interesting and powerful explanation I have heard is that the near-monoculture is the result of participation by most of the professional media (as well as much of academia) in what is described in game theory literature as a coordination game. An example of a coordination game is that, even though there is no inherent reason why one side of the road is better than the other, whichever side your neighbors drive on is the side that you want to drive on.
Shannon Love described the news media as a "Parliament of Clocks." A hypothetical clock buyer may have no basis for judging the accuracy of a clock except by comparing it to other clocks. The clock salesman is thus likely to be more concerned with making sure that his clocks are consistent with his competitors' clocks than with making them accurate. In this case, a clock's accuracy is what is known in economic literature as a "credence good," and the clock sellers in Love's example promote credence by creating an illusion of consensus. Laurence Iannaccone discusses this in the context of economics of religion (Risk, Rationality, and Religious Portfolios) [moved behind paywall]. My short version is here.
But what exactly is this credence good, and who is the audience for whom the illusion of consensus is being created? Love's clock example illustrates an illusion of consensus being used by insiders to sell a credence good to outsiders. But Fox News and conservative AM talk radio are doing fine financially by trashing the consensus and catering to a historically underserved market. It's not clear that conforming to the mainstream consensus is actually profitable. A causual reading of a typical newspaper article on any subject on which the reader is familiar (e.g. what airplanes were seen at an airshow) is usually enough to dispell any belief that newspapers are driven by a burning desire for accuracy. There have also been far too many incidents like "Rathergate" for a reasonable person to believe that the mainstream media are dedicated to fairness.
Instead, in so far as the mainstream media are producing a credence good, it seems to me that, like churches, they produce their credence goods largely for their own consumption. This makes sense if (1) the media have historically been somewhat sheltered from market forces, (2) their employees are motivated by career satisfaction more than by abstract principles of fairness, and (3) employees' career satisfaction is driven by collective belief in a flattering worldview that is functionally equivalent to the supernatural beliefs taught in churches. Lee Harris describes a friend with such a worldview:
What I saw as a political act was not, for my friend, any such thing. It was not aimed at altering the minds of other people or persuading them to act differently. Its whole point was what it did for him.
And what it did for him was to provide him with a fantasy — a fantasy, namely, of taking part in the revolutionary struggle of the oppressed against their oppressors. By participating in a violent anti-war demonstration, he was in no sense aiming at coercing conformity with his view — for that would still have been a political objective. Instead, he took his part in order to confirm his ideological fantasy of marching on the right side of history, of feeling himself among the elect few who stood with the angels of historical inevitability.
Once a media institution has been captured by the "mainstream," it tends to stay captured. Part of the reason for this is that the employees of a captured institution resemble the parishioners of a church in that they have been self-selected to be comfortable with the institutions' worldview. New employees come to the company already in a gleichgeschaltet condition, perhaps in part from having been trained in a journalism school that was itself long ago captured. Once in the business, both career satisfaction and the subjectivity of the existing hierarchy in awarding promotions tend to drive out or marginalize the few heretics and apostates. As with churches, the presence of apostates within an institution poisons the illusion of consensus. Apostates like John Stossel and Lou Dobbs are therefore shunned by orthodox members of the mainstream, including formal and informal professional societies, if not forcibly driven out by the institution's authorities. The mainstream institutions are only threatened by the restoration of market forces, such as the termination of the "Fairness Doctrine" and new competitors that result from technological changes or come from outside the mainstream.
The claim that media behavior is driven by employee career satisfaction is widespread. Ed Driscoll quotes Glenn Reynolds:
"As I keep saying over and over again, the 'killer app' for Big Media is hard news, accurately reported. That seems like something that they resist. It's almost like their position is that they didn't go into the news business to report facts accurately," he chuckles; "that's boring!"
Mencius Moldbug quotes from Walter Lippmann's book, Public Opinion to the same effect:
Reporting, which theoretically constitutes the foundation of the whole institution, is the most poorly paid branch of newspaper work, and is the least regarded. By and large, able men go into it only by necessity or for experience, and with the definite intention of being graduated as soon as possible. For straight reporting is not a career that offers many great rewards.
Update, 2-5-2011: Radley Balko says "The Media Aren't Liberal." They're statist. They're statist even on issues like drug policy and eminent domain where the American left is sympathetic to libertarians.
Update, 8-4-2013: See Moldbug's explanation of The Cathedral in Open Letter part 4. There are three ways to acheive a consensus: 1. coercion, 2. spontaneous attraction (the intellectual equivalent of a bicyclists' peloton), and 3. reproducible evidence. Moldbug's "Cathedral" is an intellectual peloton. Read the whole thing.