January 30, 2005

Larry Summers and Women Scientists--Posner

Larry Summers, the president of Harvard, stirred up a hornet’s nest when, at a recent conference on the underrepresentation of women and members of minority groups in science and engineering, he suggested the following two possible reasons why women are underrepresented. First, women’s math and science aptitude test scores exhibit less variance than men’s and this difference may have a biological basis. Second, women are on average unwilling to make the same sacrifice of time to career that men are willing to do. (A third reason, he suggested, might be discrimination against women.) Conference. (For an interesting discussion of the issues, see Saletan.) I want to consider whether there is any merit to his suggestions—but also whether he should have raised the issue at all, given his position as the president of the nation’s best-known university, and whether, having done so and been criticized, he should have apologized, as he did; he said that he had been “wrong to have spoken in a way that has resulted in an unintended signal of discouragement to talented girls and women,” although he did not repudiate the content of his remarks. Summers.

Were Summers an expert on the reasons for gender-related occupational patterns, and as a result had special insight into the issue of women’s lack of proportional representation in science careers, there might have been a real cost in his failing to speak to the issue. However, since he is not an expert in this area, there would have been no great loss to human knowledge had he kept silent and let the experts engage with the issue. Although it is a highly sensitive issue, it is not—unlike the issue of racial differences—so hot a topic that no reputable academic dares investigate it.

So the benefit of Summers’s speaking out was small. The cost would have been small, too—were he not the highly visible president of the nation’s most famous university. For as a practical matter, chief executive officers do not enjoy freedom of speech. A CEO is the fiduciary of his organization, and his duty is to speak publicly only in ways that are helpful to the organization. Not that he should lie; but he must avoid discussing matters as to which his honestly stated views would harm the organization. (Judges also lack complete freedom of speech; as I mentioned in our introductory blog posting, I am not permitted to comment publicly on any pending or impending court case.) Summers must think that his remarks did harm the university, as otherwise he would not have apologized—for he apologized not for what he said, but for saying it.

A university president might make provocative remarks because he wanted to change his university in some way, for example by encouraging greater intellectual diversity, or because he wanted to signal strength, independence, intransigence, or other qualities that he thought would increase his authority, or even because he wanted to intimidate certain faculty by seeming to be a “wild man.” But that explanation is not available to Summers, because of the apology. And the apology was probably another error, whether or not he should have raised the issue of women’s relative scientific aptitudes or tastes in the first place. The apology signaled weakness, and it cannot help a leader to appear weak. Summers has enemies in the Harvard faculty who will be encouraged by his apology to press him for concessions on issues important to them—such as diversity hiring.

The apology was also condescending. It assumed that women’s career commitments are so fragile that Summers’s remarks at the conference would actually reduce the number of women who choose a science career. Science is a tough career, both highly competitive and not very well paid. It is not for the fainthearted of either sex. If (as I doubt) women are as easily discouraged as Summers’s critics believe, their future in science is not bright.

The apology was particularly unfortunate because it dignified the criticisms of Summers’s remarks at the conference, and those criticisms were obtuse—which brings me at last to the substantive issue. The critics misunderstood Summers to have been claiming that female scientists are inferior to male scientists. Not at all. He made no comparison between male and female scientists. He was venturing possible explanations other than discrimination (the politically correct explanation) for why there are fewer female scientists than male. The ratio of female to male scientists is unrelated to the average quality of female and male scientists, and indeed is consistent with the average female scientist’s being abler than her male counterpart. In fact if, as Summers’s critics allege, and Summers admitted was a possibility, discrimination against women is a major cause of the imbalance in the number of male and female scientists, the implication is that the average female scientist is probably abler than the average male scientist. Employment discrimination usually manifests itself in a refusal to hire a person in the disfavored class unless he or she is so superior that the refusal would impose serious costs on the institution and perhaps invite a lawsuit. When anti-Semitism was rife in universities, it was assumed that a Jew had to be abler than a gentile to obtain a university appointment; it would follow that the average Jewish professor was abler than the average gentile professor in that era.

Summers’s suggestion that women on average (an essential qualification, obviously) are not as willing to invest as much time in a career as men should not have been controversial. Women who want to have children, as most do, must expect to devote more time to child care that men do. That is a brute fact and has nothing to do with scientific careers as such. Summers’s controversial conjecture was that since science-related aptitude tests exhibit less variance in female than in male scores, there are likely to be fewer women in both tails of the distribution—fewer scientific dopes but also fewer scientific geniuses. Imagine two bell curves, each with the same mean but different variance, superimposed on each other. The bell curve with the smaller variance (female) will be narrower and thus have shorter tails. So as one moves toward the end of each tail, the population with the greater variance (male) will increasingly be overrepresented. This will affect the relative number of the two populations in the tails; it may or may not affect the average quality of the members of each population who are in the tails.

Summers rightly offered the variance story as a speculation rather than as an established truth, though another fact consistent with it, besides the test scores, is that at the undergraduate level women’s science performance is equal to men’s—for at that level, one is not as far out in the tail as at the graduate level. You don’t need as much science talent to obtain a B.S. as to obtain a Ph.D.

Could the difference in variance have a biological basis? That is a legitimate subject of inquiry, which is all that Summers suggested. I cited Saletan’s article, which unlike most media coverage of the controversy engaged with the issues rather than merely playing it as a fight between angry feminists and an embattled public figure. But Saletan made one silly argument. It is that the likelihood of a biological explanation for the gender imbalance in science is enhanced by the fact that a man has more genes in common with a male chimpanzee than with a female human being. It is a surprising fact, but it may well be entirely explicable by the different biological roles of male and female in reproduction; it need have no connection to scientific aptitude.

Summers said that discrimination may also contribute to the imbalance between male and female scientists. It is certainly in the national interest to eliminate such discrimination, as he strongly believes. Nevertheless the fact that there may be nondiscriminatory reasons for disparities in occupational choice deserves investigation. Discrimination has declined, yet occupational disparities between various groups persist, suggesting that we should be looking for causes that are unrelated to discrimination as well as those that are related. A glance at the composition of different occupations shows that in many of them, particular racial, ethnic, and religious groups, along with one or the other sex and even groups defined by sexual orientation (i.e., heterosexual versus homosexual), are disproportionately present or absent. For example, a much higher percentage of biologists than of physicists are women, and at least one branch of biology, primatology, appears to be dominated by female scientists. It seems unlikely that all sex-related differences in occupational choice are due to discrimination; and therefore someone who explores alternative explanations should not be excoriated. Unless perhaps he is a university president!

Posted by posner at 06:39 PM | Comments (66) | TrackBack

Comment on Gender Differences in Scientific Achievement-BECKER

It is surely legitimate to raise the issue of biological differences in explaining the lower number of female than male scientists. But the issues are more complicated and to some extent different than the ones that are frequently stressed.

The basic question in this regard is: how much of the difference in numbers and achievements of male and female scientists is explained by biological factors compared to other factors? We would have been greatly mistaken if we concluded 40 years ago that the very small representation of women in law, medicine, business, engineering, and many other professions was mainly due to any limited aptitudes for these fields. For since then, the fraction of female students in medical, law, engineering, and business schools rose rapidly, and women are more represented than men in some very good graduate programs in these fields. Their biology did not change, but birth rates declined, and women’s education and labor force participation increased rapidly. These forces, combined with an assault on discriminatory barriers to entry in these fields, were clearly the major ones involved in the very rapid growth of women’s participation in these professions.

So what priority should be given to biological aptitude rather than time spent in child rearing, discrimination, social conditioning, and other non-biological factors in explaining the continuing under-representation of women among scientists, and even more among top scientists? No one knows for sure- which is why academic pressure against discussing possible biological difference in talents is disturbing. However, my own belief is that we can get a lot of explanatory power out of factors that do not rely on intrinsic gender difference in talents, including high-level talents.

The reasons behind this conclusion are simple. To be a top level scientist-indeed, to be tops in any challenging field- requires long hours of work and an intense commitment to discovery and the like. Yet as long as women continue to have the major responsibilities for child-rearing and other household activities, they will have to combine professional activities with a mother’s and other household duties. Inevitably, that will force most women to reduce their professional commitment.

These women will adjust either by lowering their scientific ambitions, or by electing not to enter these fields in the first place. Others will forego motherhood and even marriage to pursue their scientific careers, and some of these and a smaller fraction of the other women will become highly successful. But even without discrimination against women, the attempt to combine several quite different activities will continue to lower the fraction of top women scientists (or top CEO’s, lawyers, etc) compared to men.

The variance in the distributions of the required talents may well be greater among men than women-as suggested by Larry Summers and others- so that there are many more brilliant and very dumb men than women. Even so, one does not want to overestimate the importance of brilliance in explaining the so-far low representation of women among outstanding achievers, as measured by Nobel and other major prizes. For a large fraction of male high achievers are not brilliant-they are not an Einstein, Newton, Euler, or LaPlace, to name a few of the recognized geniuses in scientific accomplishment.

An outstanding Columbia University physicist, the late I. Rabi, years ago was supposedly asked at a gathering of Nobelists and other high achievers about how most of those present had achieved so much since they did not seem particularly brilliant. His brief answer: “hard work”. That is also my belief after being at many similar gatherings.

Women are likely to be at a much greater disadvantage in this regard, due to their child-rearing and other responsibilities, than in biological aptitude. While studies indicate that the total hours worked by women, including household “work”, are generally as high or even higher than the total hours worked by men, women’s work is less specialized toward professional and other business achievement. Moreover, they anticipate this lesser degree of specialization in determining their professional ambitions and time use at early ages.

For other reasons as well, it is difficult to infer biological differences from occupational choices. For example, biological factors could entirely explain occupational choices, and yet the lower representation of women among scientists would not imply that they have less scientific aptitude. The reason is that women could be better than men at all occupations, but would be underrepresented in science if any difference between men and women in scientific aptitudes were smaller than in non-scientific aptitudes.

In my book, A Treatise on the Family, I expressed a belief that the traditional gender division of labor between working in the marketplace and working in the household- that is, taking care of children, etc- is partly due to biological differences between men and women. However, I also stressed that this gender division of labor is consistent with women being superior to men at market activities too. Rather, it implies that differences in market “abilities” are less than at child rearing and the like. In economic jargon, observed data on occupational choices only reflects comparative advantage, not absolute advantage.

My conclusion is that the sharp differences in scientific and similar accomplishments between men and women may be partly due to differences in high-level aptitudes, but that such differences are less important than other forces. To be sure, scholarly studies of any biological differences between men and women should be welcomed. Still, I believe that studies of other influences on male-female differences in scientific and related achievements are likely to be highly productive.

Posted by becker at 06:01 PM | Comments (6) | TrackBack

January 28, 2005

Response to Comments on Profiling--Posner

There were as usual many interesting comments, not all of which I can reply to. (Among critics of my position, I particularly commend the thoughtful comment posted by Anonymous on January 24.) I was however startled by the large number of comments that compare profiling to affirmative action and ask that commenters who oppose profiling as demeaning, alienating, and so forth take an equal stand against affirmative action. Although I have serious reservations about many forms of affirmative action, and although there is indeed a conceptual parallel between it and profiling, since in both cases a single criterion, such as ethnicity, is used as the basis for imposing benefits and burdens respectively, the symmetry is incomplete. The reason is simply that most beneficiaries of affirmative action are happy to have the benefits! Most people take for granted whatever advantages they have, however adventitious and undeserved. What is more adventitious than having wealthy parents? And yet how many rich kids are bitter because they have been singled out for benefits unrelated to their merit?

My argument against racial reparations, and likewise compensation for victims of profiling, is not that the beneficiaries will lose self-esteem or otherwise be immiserated by being benefited, but that using ethnic or racial or other such criteria for benefits treats the benefited group as being importantly different from the rest of the community. I would think it healthy for Americans to become less conscious of their differences, whether the differences are based on race, sex, national origin, ethnicity, politics, or sexual orientation, and think of themselves, rather, as being "just Americans." That would certainly help in presenting a united front against the threat, which is real and probably growing, of international terrorism. It is particularly important that Arab-Americans and other Americans of Middle Eastern origins or Muslim religion feel fully American. At the risk of seeming an alarmist (a "McCarthyite," some might call me), I believe that there are almost certainly al Qaeda sleeper cells in the United States, and it is extremely important that they not receive any assistance, financial or otherwise, or protection, active or passive, from the Muslim community. Cementing the community's loyalty to the United States is a vital national project, and this has to affect the amount of profiling that is in the national interest.

I do not agree with the comment that, in defense of remedial affirmative action, describes profiling as a product of "entrenched bias." When profiling is based on a relevant characteristic, such as a known greater propensity to engage in some antisocial behavior, it need have no connection whatsoever with bias, entrenched or otherwise.

A comment I strongly disagree with is that profiling airline passengers is unsound because no terrorist has ever been intercepted as the result of profiling. First, we don't know whether this is accurate; people are occasionally stopped from boarding a plane because of a secondary search prompted by their profile, and some of these people may be terrorists though they cannot be proved to be. Second and more important, knowing that there is profiling may discourage some terrorists from attempting to board an aircraft, since if they are arrested their career as terrorists may be terminated before they can do any harm.

Posted by posner at 08:49 PM | Comments (13) | TrackBack

January 27, 2005

Reply on Profiling-BECKER

Let me reply briefly to the main comments.
One theme in the negative comments is that if one profiles terrorists, or groups believed to be more likely to commit other crimes, the terrorists or criminals simply change whom they use, or new criminals replace those who are profiled. Sometimes that does defeat profiling, but not usually. Terrorists cannot change that easily whom they use since the supply of types willing to give up their lives to kill Americans or others is highly limited. Similarly, the supply of criminals typically has a limited elasticity of response. These considerations explain why the police "profile" ex-convicts, young people congregating at various suspicious locations, and in other ways. Are the critics saying the police should not be doing this? I doubt if they would carry that position through consistently.

Yes, we can spend more on various types of deterrence, but do we really want to search elderly women at airports as much as young men of Middle Eastern origin? (Let me add that my wife and her brothers are of such origin, and accept that they do get profiled.)

I recognize that being singled out for extra attention can be embarrassing. I suggested compensation for that. Someone mentioned that Steve Landsburg has also suggested compensation-good for Steve! Such compensation may not be sufficient to offset fully the damage from being singled out, but along with courteous and respectful treatment, it can go a good ways toward reducing the distastefulness of the process.

One last point. I work weekends in my university office when the doors are locked and few persons are in the building. I "profile" in judging whether to call the campus police by whether I believe a person belongs in the building- in fact I have called the police on more than one occasion. Do the critics of profiling suggest that instead of such profiling I call the police every time I see someone in the building-that is the purest example of no profiling- or every time I see someone in the building that I do not recognize, or when I judge they do not belong in the building, even though my judgment may depend on age, race, or other group characteristics?

This is not only costly, but borders on the absurd. I believe that the commentators who claim to oppose profiling do in fact profile in this and thousands of other situations. I do not know if this is true, but the media reported a few years ago that Jesse Jackson profiled when he heard someone walking behind him, and turned around to determine what group they belonged to!

Posted by becker at 07:18 PM | Comments (12) | TrackBack

January 23, 2005

Response on Torts-BECKER

As usual, I wish I had some of these comments before I wrote mine! Just a few reactions.

Yes, the Macdonald's case is more complicated, but no one made a convincing case that the customer could not have tasted the coffee first to determine how hot it was. Macdonald's criteria in serving hot coffee are not the real issue if it were easy for customers to check the nature of the product they consume. That surely was true in this case.

One can separate, at least to some extent, who gets punitive damages from who gets the compensatory damages. But one also wants to give victims sufficiently strong incentives to sue.

I generally do support allowing customers and providers of medical care, other services, or goods, to reach contracts ahead of time about rights to sue, compulsory arbitration, and so forth. Still, the same considerations that justify lawsuits-lies, withholding relevant information, and terrible service- would also be relevant in interpreting such contracts.

Posted by becker at 08:36 PM | Comments (9) | TrackBack

Is Ethnic and other Profiling Discrimination? BECKER

The value of using group membership in judging unobserved characteristics is uncontroversial most of the time, and so is hardly noticed. For example, automobile insurance companies consider young unmarried males as a relevant group in determining driver insurance premiums because they tend to have more car accidents than older males or young women. These higher insurance rates also help cut down the number of auto accidents by reducing driving by accident-prone young males. Yet given that group membership is almost always an imperfect predictor of unobserved characteristics, some individuals will be treated much worse (or better) than their true characteristics justifies. In the driving case, young unmarried males who are careful and responsible drivers will pay more for insurance than they would in a world with better information. They might be discouraged from driving because they suffer from the bad driving of other young unmarried males.

Still, limited information about individuals means that group identities is often useful in gaining information about them. However, the use of religious, racial, or ethnic characteristics for national security protection and in fighting crime has been a political hot potato, and has led to accusations of discrimination on the basis of race and other characteristics. For the use of group identities in order to treat different groups differently may be the result of a desire to discriminate against various minority groups instead of a desire to act effectively to prevent some undesirable actions.

On top of the current agenda is the effort by organizations fighting terrorism to treat Muslim with greater suspicion. Is it justified to single out young Moslem males from the Middle East for much more careful searches at airports, or for tougher requirements to gain tourist visas and green cards? Or are employees at stores that are trying to prevent theft of their goods justified in watching minority customers more carefully than they watch others? Macy’s was recently fined for allegedly watching blacks and Hispanics more carefully, although the company denies that such profiling of customers is their policy.

Efficient actions would say minimize the amount of terrorist activities in the US, given a limited amount to be spent on preventing terrorism, or would minimize store theft, given a budget for security personnel. Then security checks at airports would try to both catch terrorists before they board a plane, and to discourage future airport terrorism by raising the chances that they are caught at the airport. Similarly, store security both tries to apprehend thieves, and to discourage future store theft.

If young Moslem Middle Eastern males were in fact much more likely to commit terrorism against U.S. than were other groups, putting them through tighter security clearance would reduce current airport terrorism. Whether such religious and ethnic profiling furthers the second goal, of deterring future terrorism, depends on the degree of responsiveness of different potential terrorists to a greater likelihood of being caught. If the degree of response by different groups were proportional to their average propensity to engage in terrorism, then checking young Moslem Middle Eastern males more carefully would also help deter future terrorism at airports.

On the other hand, profiling by average propensities might be inefficient if the marginal propensity to reduce terrorism with more careful checks were smaller for groups like Moslems that might have higher average propensities. That could well be true if these groups were more fanatical and less easily deterred by the prospects of being caught. Then the “deterrent” effect on future terrorism would be opposite to and might be stronger than the “apprehension” effect on current terrorism.

The deterrent effect is less likely to dominate the apprehension effect when the difference in average behavior is greater. This is why it is efficient to profile young male Middle Eastern Moslems for terrorist attacks at airports, and perhaps it is efficient also to watch minority shoppers more carefully at stores. To be sure, such profiling is “unfair” to the many young male Moslems who are not terrorists, and to the many minority shoppers who are honest. That could be made up in part by compensating groups who are forced to go through more careful airport screening through putting them in shorter security lines, or in other ways. Similarly, innocent shoppers who are stopped and searched could be compensated for their embarrassment and time.

To be sure, some profiling by governments and the private sector has been due to prejudice against various groups, not as a way of achieving efficiency. So it is crucial to be able to distinguish whether a profiling is efficient from whether it is evidence of discrimination. This distinction can be made in the terrorist field (similar considerations apply to fighting crime) by keeping records on the fractions of young Moslem males and others who were searched and found with weapons or other evidence of intent to commit a terrorist act. If the fraction were much greater among Moslems searched than among others, this would at least be consistent with an emphasis on efficiency rather than discrimination.

A further test would be to determine what happens to apprehension rates as the amount spent on airport security increased or decreased. The profiling policy would again be consistent with efficiency if greater spending on airport security reduced the apprehension rates of young male Moslems who were searched about as much as the apprehension rates of others.

So it is possible to provide analytical criteria and guidelines to determine when particular types of profiling are explained by efficiency considerations rather than discrimination. By using these guidelines to analyze data on apprehension rates, one can determine in an objective manner whether discrimination rather than efficiency is responsible for different treatment of members of ethnic, religious, or racial groups.

Posted by becker at 08:27 PM | Comments (18) | TrackBack

Comment on Profiling--Posner

At one level, “profiling” is unexceptionable. If witnesses report a theft by a young black male, it would be absurd for the police to look for suspects among other groups in the population. Profiling becomes interesting only when the differential probability of “guilt” is much smaller. Even then, it is unproblematic, as Becker notes, when the disfavored group is not a sensitive minority. No one objects when smokers are charged a higher price for life insurance than nonsmokers, even though many smokers outlive many nonsmokers. Even when the condition that puts one in the disfavored class is involuntary, such as having a family history of cancer or heart disease, “discrimination” on this basis (what economists call “statistical discrimination”) is generally considered permissible because it is not exploitive or based on hostility or contempt and it does promote a more efficient allocation of resources.

Profiling based on race, sex, or national origin, however, is intensely controversial. It is helpful in discussing it to make two distinctions: between ordinary crimes and Islamist terrorism (e.g., al Qaeda), and in the terrorist case between profiling U.S. citizens and profiling foreigners. I will be discussing these issues purely as issues of policy rather than of law.

In the case of ordinary crimes, where for example profiling might take the form of disproportionately frequent searches of vehicles driven by Hispanics because Hispanics are disproportionately represented in illegal drug trafficking, I would expect profiling to have little effect on the crime rate. The reason is the positive elasticity of supply of persons who commit victimless crimes, which is to say crimes resulting from the outlawing of products or services for which there is a demand. If one class of suppliers is driven out of business, this makes room for others. Given the fixed budget for law enforcement assumed by Becker, the increased apprehension of Hispanic drug couriers would be offset by a reduced risk to non-Hispanics of being apprehended for transporting drugs, and so the non-Hispanics would flock to replace the Hispanics as couriers. The ethnic composition of the illegal work force would be altered by profiling, but the crime rate would be affected only to the extent that Hispanics are more efficient drug couriers because of language and other ties to major drug supply countries; the net effect on the crime rate would probably be small.

In the case of terrorism, a similar replacement effect can be anticipated, although it would probably be smaller. Assume a fixed budget for screening airline passengers and a reallocation of funds within the budget limit to enable more young male airline passengers who appear to be Muslim (or of Middle Eastern origin, but for simplicity I’ll assume that Muslim-appearing is the screening criterion) to be subjected to intensive screening, as distinguished from the limited screening to which all passengers are subjected. Then fewer passengers who do not fit the profile will be screened (this is implied by the fixed budget), which will induce terrorist groups to make greater use of female Muslims (as happened in suicide attacks in Israel), older Muslims, and young Muslims who do not appear to be Muslim, for members of these groups will now be less likely to be apprehended than before the adoption of profiling. The elasticity of supply of terrorists is probably not as great as that of drug couriers, but it is positive and will reduce the effect of profiling. A parallel analysis recommends against concentrating too many of our antiterrorist resources on the protection of New York and Washington, since terrorists can substitute other targets.

The benefits of airline passenger profiling are thus likely to be modest, and the costs may be great in the case of Muslims who are U.S. citizens. Being singled out on the basis of race, religion, or ethnic origin is intensely resented by the people who are discriminated against and could undermine their loyalty to the United States if they have strong ethnic and religious ties with the nation’s enemies. A paramount goal of U.S. antiterrorist policy should be to prevent the disaffection of U.S. citizens of the Muslim faith and Middle Eastern ethnicity. That goal would be undermined by profiling. I do not think compensating them financially for the additional inconvenience would rectify the problem; indeed, it would underscore their differentness from their fellow citizens. (This is also an argument against reparations for blacks and American Indians.)

The argument for the efficiency of profiling is further undermined by relaxing Becker’s assumption of a fixed security budget. By increasing the budget for airline security, it would become possible to screen everybody carefully. I suspect that the optimal policy is to subject more U.S. citizens of apparent Middle Eastern origin or Muslim religious identity to intensive screening than other citizens, but to subject enough of the other citizens to the same intensive screening so that the (lightly) profiled group does not feel markedly discriminated against--and so that substitution of terrorists who do not fit the profile is held in check.

My view with regard to profiling noncitizens is different. Noncitizens are not expected to be loyal to the United States and so the concern with alienating them by profiling is less acute. No foreigner expects to be treated identically to a citizen.

Posted by posner at 08:05 PM | Comments (32) | TrackBack

Tort Reform--Posner's Reply to Comments

As usual, there were many excellent comments. Let me respond to some:

Several comments point out correctly that the determination of medical malpractice (that is, medical negligence) by the courts is very often inaccurate; there are many false positives and false negatives. To that problem, capping damages judgments is no solution. An attractive solution is testimony by a neutral expert witness. The fact that judges may have difficulty determining who is neutral is no objection; the judge can ask the parties' medical experts to jointly nominate a third; he would be the neutral and the judge and jury would appropriately rely heavily on his testimony. The procedure I am suggesting is similar to a widely used procedure for picking a neutral arbitrator: each party designates one arbitrator, the two arbitrators choose the third, who is neutral, and he then provides the deciding vote.

An alternative, mentioned in one comment and already in force in a number of states, is to require the malpractice plaintiff before suing to submit his claim to a panel of physicians, whose findings, if unanimous, are admissible in court should the claim result in a lawsuit.

One comment pointed out that medical errors are often systemic, i.e., they result from erroneous procedures or practices by hospitals, drug companies, and other institutions rather than from mistakes by individual physicians. However, those entities are suable.

It was also noted that heavy insurance premiums might drive some physicians from practice and deter some people from becoming physicians in the first place. That is true, but if the result is less medical negligence, the benefits might exceed the costs. In addition, the overall effect on medical expenses is likely to be slight, because physicians' fees are only a moderate component of overall medical expenses. Furthermore, if physicians are driven out by high premiums, the resulting reduction in the supply of physicians should enable those who remain to raise their fees.

A slightly esoteric point: one comment suggested that pain and suffering, disfigurement, and other nonpecuniary losses imposed by medical errors are not real costs because people rarely try to buy insurance against such losses. However, the reason they do not buy insurance is not that the losses aren't real, but that insurance is designed primarily for replacing income or defraying an expense.

I also disagree that negotiation of the level of medical care should be left to physician and patient, because they have a preexisting contractual relationship. The principle is fine (though it would require a chance in existing law), but the transaction costs would be prohibitive because of the patient's ignorance of particular procedures, risks, and so forth. In addition, a physician who told his patient that he would operate on him only if he waived his right to sue for medical malpractice would be signaling the likelihood of an unfavorable outcome. Hence physicians would be reluctant to suggest such waivers.

I was pleased to learn from two of the comments that some insurance conmpanies do experience-rate medical malpractice insurance. Why others do not is a mystery, but it occurs to me that one possibility is that the inaccuracy of judicial determinations of malpractice is so great that being sued and losing a malpractice case does not provide useful information about the likelihood of being sued in the future. On this view, malpractice liability is random. One hopes not; but if so, reforms, such as those suggested above, aimed at increasing the accuracy of malpractice determinations are urgently needed.

Posted by posner at 02:33 PM | Comments (3) | TrackBack

January 16, 2005

Tort Reform--Posner

There is a movement afoot, assisted by the strengthening of Republican control over Congress, to impose federal limits on tort litigation, particularly medical malpractice; premiums for malpractice insurance have soared in the last two years and physicians are protesting vigorously.

The costs of malpractice premiums are only about 1 percent of total U.S. health-care costs. Moreover, insofar as physicians are forced to swallow the cost of the premiums rather than being able to pass them on to their patients or their patients’ insurers in the form of higher prices, the premiums do not actually increase total health-care costs. There is an indirect effect, however, insofar as malpractice liability causes doctors to practice defensive medicine. But there may be offsetting benefits, to the extent that defensive medicine actually improves outcomes for patients; and surely it does for at least some. What is more, because malpractice insurance is not experience-rated—physicians are not charged premiums based on their personal liability experience—malpractice liability may have only a slight effect on physicians’ methods or carefulness, except insofar as physicians are pressured by their insurers to change their methods in order to reduce the amount of malpractice litigation.

The relation between malpractice premiums and malpractice judgments is also uncertain. No doubt capping judgments, which is the principal reform that is advocated, has some tendency to reduce premiums, but perhaps not much, because there is evidence that premiums are strongly influenced by the performance of the insurance companies’ investment portfolios.

A better reform would be to permit, encourage, or even require insurance companies to base malpractice premiums on the experience of the insured physician, much as automobile liability insurance is based on the driver’s experience of accidents. That would make malpractice liability a better engine for deterring malpractice—which in turn would reduce malpractice premiums by reducing the amount of malpractice. Capping judgments, in contrast, would reduce the incentive of insurance companies and their regulators to move to a system of experience-rated malpractice insurance.

It is always important to distinguish between financial and real costs. Insofar as malpractice liability merely transfers wealth from physicians to (some) patients, aggregate costs are unaffected. The real cost of malpractice liability is limited to the cost of the actual resources consumed by such liability, principally the time of lawyers and expert witnesses (roughly half the total amount awarded in judgments goes to pay lawyers and expert witnesses), unless defensive medicine is assumed to cost more than its benefits in improving treatment outcomes. The real benefit of malpractice liability is its effect if any in deterring medical negligence; reducing that benefit would impose a real cost. Hence it is simplistic to assume that the total annual malpractice premiums paid is a good index of the net social cost of malpractice liability, or that measures to reduce those premiums by capping malpractice liability would result in a net improvement in welfare. To repeat, part of the premiums represent simply a wealth transfer from physicians to the patients who receive malpractice judgments or settlements paid by insurers. The part (roughly half) that pays for lawyers and expert witnesses should be understood as the cost of maintaining a system for increasing medical safety; the efficacy of the system could be improved, I have argued, by experience rating, but not by capping judgments.

In any event, there is no compelling case for federal limitations on malpractice liability. The issue belongs at the state level, and as reported in a New York Times article last Friday, a number of states have adopted or are seriously considering adopting the kind of caps being advocated in Congress. Federal legislation would simply stifle state experimentation with different methods of regulating physicians and prevent us from learning which is best.

There is a stronger case for federal regulation of class actions, as in the case of suits against asbestos manufacturers. When the members of a plaintiff class are scattered across the country, the class lawyer has a wide range of places in which to sue, and there are certain counties in the United States in which judges and juries are disproportionately generous to tort plaintiffs. Most of the costs of a large judgment or settlement in such a case are exported to other states, while the benefits are concentrated in the locale where the suit was litigated, because of the business generated for local lawyers, as well as the judgments or settlements received by the members of the class in the locale. This is a formula for abuse, concretely for a tendency for such judgments and settlements to exceed an unbiased estimate of the true costs imposed on the class by the defendants’ misconduct. Malpractice litigation does not give rise to such an abuse to any very great extent, because patient and physician are usually in the same state, and a single plaintiff has only a limited choice of courts in which to sue. This is another reason not to make medical malpractice the principal object of federal tort reform.

We should be cautious about tort reform. It would be unfortunate if interest-group politics, and anecdotes concerning outlandish lawsuits (such as the suit against McDonald’s by the customer who spilled hot coffee in her lap), were allowed to obscure the difficult policy issues.

Posted by posner at 08:59 PM | Comments (41) | TrackBack

Comment on Tort Reform-BECKER

I agree with Richard Posner that the ability to sue against medical malpractice, drunk drivers, and sellers of defective and dangerous products induces doctors, producers, and others to pay closer attention to what they do. I also agree that reforms are generally better left to the states than to the federal government in order to allow competition among state legal rules. However, without reform of class action suits, Posner’s appropriate criticism of the right to bring class suits in more generous jurisdictions undermines the argument to leave legal reforms entirely to the states.

He is right too that a few foolish cases should not be the basis for policy. Yet there is need for some tort reform because there is too much litigation, compensatory damages are often too low, and punitive damages have sometimes been grossly excessive.

“Compensatory” damages are supposed to compensate individuals for their losses from medical malpractice, and other careless or reckless behavior. In practice, they commonly refer to the loss of earnings due to injury or death from such actions. Yet the correct measure of compensatory damages should equal what individuals are willing to pay to avoid death or the injuries in question. The value placed on loss of life, or on substantial disabilities, are usually many times greater than the loss in earnings.

Critics of the American tort system generally neglect this tendency to underestimate compensatory damages. To be sure, this is not easily corrected by legislation, although it might be possible to have useful rules of thumb about the appropriate ratio of full damages to lost earnings. And greater recognition that compensatory damages are often much too low would be helpful to juries and judges.

“Punitive” damages are often imposed in order to encourage doctors, drivers, and producers to be more careful and responsible. The system would be working well if any underestimated compensatory damages were just about offset by inflated punitive damages. But punitive damages have sometimes been huge multiples of compensatory damages -occasionally exceeding 50 or 100 times compensatory damages- and may only reflect a jury’s or judge’s perception about how deep are the pockets of defendants. Although huge punitive damages tend to be reduced on appeal, they often remain much too large.

The rule of thumb in anti-trust litigation is that punitive damages equal to three times compensatory damages sufficiently recognizes that many anti-trust violators are undetected and are not punished. A limited ratio of punitive to compensatory damages would also be appropriate in the great majority of other tort cases.

The Supreme Court in a couple of cases in recent years has in fact suggested that punitive damages should generally not exceed nine times (I would add correctly measured) compensatory damages, and their ruling applies to state as well as federal punitive damages. I believe that as in anti-trust cases, the ratio should typically be much less than 9. One exception, first articulated in an opinion by Judge Posner, is when compensatory damages to each person harmed are very low, while the total harm may be large because many persons suffer small loses. In these cases, it would be desirable to raise the punitive-compensatory ratio to encourage lawsuits to discourage the behavior that produced these harms.

Excessive litigation in the United States creates inefficient defensive practices by doctors and others, and sometimes destroys good companies, as in the asbestos litigation, and perhaps now in the looming Vioxx class action suits against Merck. Litigation against doctors and others is justified when they could have but did not take sufficient care to insure safety, if they lied, or if they withheld evidence about safety. In these cases, doctors and producers should be responsible because they have much easier access to the relevant information than do patients and consumers.

On the other hand, consumers should bear responsibility for their actions when they could have easily prevented the damage. The MacDonald’s customer who drove with hot coffee on her lap should not have won her lawsuit, and perhaps should have been forced to pay at least some of MacDonald’s legal expenses. Similarly, parents who carelessly allow their children access to dangerous medicines should be responsible, not the manufacturer of medicine bottles.

In many cases there is reasonable doubt as to whether patients or doctors, consumers or producers, could more readily have avoided the harm. I believe in these cases there should be a presumption toward “caveat emptor”- buyer beware. A bias toward buyer beware when there is considerable doubt about who could have more readily avoided the damages would cut out much useless and costly litigation without discouraging the justified cases where there is negligence, fraud, or information suppression by doctors, drivers, producers, and others.

Posted by becker at 07:36 PM | Comments (17) | TrackBack

Reply to Student-Loan Comments--Posner

There were as always interesting comments, butI will reply just to the one that was made most frequently. It is that there are external benefits from the study of certain subjects in college, such as English literature and other of the humanities (philosophy, classics, etc.), that would be lost without subsidized student loans. I disagree with this argument for subsidized loans, on several grounds. First, while there may be some external benefits to the study of the humanities, although I'd like to see the evidence or even the argument, I imagine that they are exhausted at a lower level than is generated by subsidization. Second, universities have flexibility with their endowments and they can allocate more endowment funds to the humanities if student tuition income drops because of withdrawal of loan subsidies. Third and most important, most of the ablest students go on to some form of graduate education, and for them the choice of an undergraduate major has few economic implications and so is unlikely to be affected by the existence of loan subsidies. One of the commenters pointed out that I myself majored in English in college. I think it was a valuable background to my subsequent legal studies and I would continue to recommend it as a college major.

Posted by posner at 06:00 PM | Comments (2) | TrackBack

Reply to Comments on Student Loans-BECKER

I can be brief since the discussion among those commenting already answered many of the issues raised in other comments.

Some claim that a free market for student loans could not work well because lenders would charge high interest rates, like credit card companies do. But high credit card rates are mainly due to the high cost of servicing these accounts because they have relatively low dollar amounts per account, and high default rates. These forces raising market interest rates would be absent under my proposed student loan program because default rates on student loans would be much lower, and the typical amounts borrowed to finance education would greatly exceed the typical debt on a credit card.

Some of the comments reflect a concern that a free market in loans would place history and other humanity fields at a major disadvantage since earnings in these fields are below those in engineering, business, and other specialties. Of course, they are already at a disadvantage under the present student loan system. The best way to encourage these fields is not by continuing to subsidize all student loans, but rather directly to subsidize students who major in humanities with lower tuition, more easily available scholarships, and in related ways. If a subsidy is desirable for any activity, it is always more efficient directly to subsidize the activity rather than using roundabout approaches.

There were several allegations of declining economic benefits of higher education. In fact, during the past three decades these benefits have gone sharply up, not down, and unemployment rates of college graduates remain way below those of high school drop-outs and high school graduates.

Fannie Mae and Freddie Mac charge substantial amounts for their insurance of home mortgages-the current attacks on these companies allege among other things that they charge too much because of their monopoly power. But the default rate is low on insured and other mortgages mainly because borrowers have to use their homes as collateral. Since student borrowers in a free market for student loans could not use their education as collateral, I proposed to offset that defect with other advantages. Perhaps most important is to continue the present system that does not allow student loans to be discharged through declaring personal bankruptcy.

Posted by becker at 02:00 AM | Comments (3) | TrackBack

January 09, 2005

Government's Role in Student Loans-BECKER

A young couple with modest earnings can buy a house that costs many times their combined incomes without putting down more than a small fraction of the cost, as long as they have reasonably stable employment and a decent credit history. They pay a market-determined interest rate to the lender that is presently about 6 per cent in the United States for a 30 year fixed rate loan. However, it is crucial that they offer their home as collateral, so that the lender can repossess it if they fail to meet their mortgage payments. They can borrow to purchase a car too, essentially by offering the car as collateral.

Yet with the exception of medical and a few other types of students, young persons cannot borrow on a purely commercial basis without parental or other co-signers to finance their higher education. The usual explanation has been that unlike homes and other consumer goods, education cannot be offered as collateral to guarantee repayment. This means that in the event of defaults on student loans, there would be no assets for lenders to repossess. Put differently, lenders cannot take ownership of the human capital they finance since that means taking ownership of the individuals receiving the education, and no modern country allows people or institutions to own other individuals. Still, even without such collateral, it is possible to encourage a fully commercial student loan market with a few straightforward changes.

Countries typically help students by paying most of the direct costs of higher education, and also sometimes by providing scholarships to help pay for the cost of living while in school. Though the stated reason for this is to allow poorer students access to higher education, the beneficiaries are mainly children from the middle and upper classes since they are the ones who attend these mainly state-run colleges and universities. Even in the Unites States, about 80% of all students are enrolled in subsidized state-run institutions. This system penalizes less prosperous families who do not have children at universities and colleges, but are nevertheless taxed to help support the education of children from richer families. Even children from modest backgrounds who do manage to get a university education tend to become part of the economic elite.

In order to provide a better way to support students, or at least a way that complements direct subsidies, the United States about 40 years ago introduced a system of loans for college students that would be repaid gradually after they finish their schooling and are working. Similar systems have been introduced in a few other nations. To encourage private lending, the federal government guarantees repayment of defaulted principal and interest. The program has grown rapidly, so that the value of student loans outstanding has increased to over $400 billion. The current interest rate on new loans is about 31/2% (it is regulated and tied to short term interest rates). Interest rates on student loans are well below rates on mortgages, credit cards, and other sources of consumer credit mainly because the federal government rather than lending institutions bears the burden of defaults.

By the early 1990’s this burden became great indeed, as defaults on student loans ballooned to over 20 per cent. No commercial loan system would be viable with these default rates at the low interest rates allowed. Partly because wealthy lawyers and doctors allegedly were responsible for significant numbers of these defaults, politicians pushed for tougher enforcement. A 1998 law made it very difficult to discharge student loans by declaring personal bankruptcy- a privilege not available to credit card companies and many other lenders to consumers- the federal government began aggressively using private agencies to collect defaulted interest and principal on student loans, and these agencies could garnish significant fractions of the earnings of borrowers in arrears on their debt. As a result, default rates have plummeted to about 5%. They are in fact relatively low among graduates of professional schools and elite colleges, and are highest- among persons who had attended community colleges and trade schools.

Loans to finance education (and other investments in human capital) are a good idea, but why should the federal government subsidize them? College education raises earnings and improves health and many other aspects of living. Since these effects are huge and even larger than a few decades ago, loans to finance education should be attractive even without subsidies. It is especially inappropriate to subsidize loans to students coming from middle class and wealthy families who can expect very high earning over their lifetimes. Some writers lament the burden of large student loans, but even $100,000 in student loans is easier to repay than $150,000 in mortgages, a burden that many young families are clearly willing to bear.

Private lending institutions would be willing to make student loans without government subsidies at interest rates that I believe would not be much higher than mortgage rates if certain conditions were met that already apply to government subsidized student loans. The most important are that commercial student loans should be hard to discharge by declaring personal bankruptcy, that collectors could garnish a reasonable share of earnings in the event of defaults, and that interest on student loans could be deducted from taxable income.

Originators of student loans-Sallie Mae and another private institutions-typically repackage them, and sell them in a secondary market, similar to the way mortgages are repackaged and sold. So the “securitization” of student loans is a reality. What is needed to have a well functioning and extensive private commercial market in student loans are the various privileges already available to government-backed student loans that help overcome the absence of collateral when making loans for education. In addition, since education is an investment in human capital, the amount students invest- perhaps measured to start by the amount they borrowed to finance their education- should also be deductible over time from their taxable incomes, the way corporations can deduct the cost of capital from their tax liabilities.

A private commercial market in student loans under these conditions would not need government subsidies, and would have the flexibility to develop new contractual forms. These might include the often-suggested system whereby interest rates on student loans would not be fixed, but would rise with the eventual earnings of borrowers. Other innovations would emerge as freer and more imaginative competition in the market for student loans developed. It is surely time to rethink the conclusion by economists that commercial loans on education are not feasible because education capital cannot be used as collateral.

Posted by becker at 07:42 PM | Comments (31) | TrackBack

Comment on Student Loans--Posner

A critical question in deciding whether the government should subsidize student loans is what the market for college education would look like without any subsidy. An important aspect of the question is what the effect would be on college attendance: would many kids who now attend college decide not to do so because the cost would be too high? I doubt that. There would be a private loan market, albeit with higher interest rates. Because college is a good investment even after the higher future incomes that it makes possible are discounted to present value, most students who now borrow for college at the lower, federally subsidized rates probably would pay the higher rates. Others would borrow from their parents. Others would take part-time jobs during the school year or work more during summers. Still others would switch from expensive private colleges to state- or city-subsidized public colleges. Colleges would grant more scholarship aid and offer their own loan subsidies. Thus the burden of the higher interest rates would be shared among students, their parents, and the colleges, though the the colleges would try to shift some of the burden to the wealthier students, to faculty, and to donors.

If despite these responses to abolition of the federal student-loan program, many college-qualified young people decided not to go to college, the question would then be—from a strictly economic standpoint and without regard to distributive justice (which I discuss below)—whether society as a whole would be a loser from having a less well-educated population. Maybe so. Educated people are more productive, as indicated by their higher incomes, and they cannot capture the entire benefits of their higher productive value for themselves; for one thing, income is shared between the earner and the Internal Revenue Service, and the latter’s share goes to benefit other people. Probably a more important consideration is that educated people are less likely to commit crimes or to end up on welfare, and more likely to vote, which may confer a benefit on the public as a whole by increasing the perceived legitimacy of election outcomes.

But the “externalities” argument for subsidizing college education depends not only on how many kids would not attend college without the federal subsidy, but also on the cost of the subsidy to the taxpayer. I have no strong sense that the net external benefits are positive. If they are positive, it is very unlikely that they are large, considering the indirectness of this method of subsidizing education.

Nor is it even clear that the subsidy transfers wealth from the better off to the worse off members of society. The subsidized loans are available to everybody, not just the needy; and the costs of the loans are paid out of federal taxes, which are no longer highly progressive.

So I am dubious about the student-loan program. But if it is retained, I see no merit in making student loans nondischargeable in bankruptcy. The effect is to reduce the interest rates on such loans, but at the same time to increase the risk to the borrower—and why is that an advantageous trade? If it is an advantageous trade, why shouldn’t consumer bankruptcy simply be abolished, so that all consumers, and not just students, can obtain the benefit of lower interest rates in exchange for losing the privilege of discharging debts by declaring bankruptcy? The benefit to borrowers of surrendering the right to discharge their debts in bankruptcy is especially illusory when one considers that someone who has a nondischargeable debt will have to pay higher interest rates to his other creditors, because if he goes broke they’ll have a reduced expectation of being able to recover any part of their loans in the bankruptcy proceeding.

This is part of a larger puzzle about secured lending. The effect of making student loans nondischargeable in bankruptcy is to make the students’ future income in effect collateral for the loans. When part of one’s property becomes security for one lender, the amount of property available for other lenders is reduced, so they charge a higher interest rate, which tends to offset the lower interest rate that one can extract from a creditor by giving him security. In the case of the home mortgage, the debt secured by the mortgage so dominates other debts for most families that the interest tradeoff clearly favors collateralization. But is that true with respect to student loans, where the collateral is income? The law does not permit a creditor who garnishes his debtor’s earnings (as the creditor of a nondischargeable debt is permitted to do) to take more than a modest percentage of those earnings. Enforcing student loans may therefore turn out to be quite costly for the lenders in relation to the benefits to them. If so, the net interest rate benefit to students from abolishing the privilege of discharging student loans in bankruptcy may be slight.

Posted by posner at 07:17 PM | Comments (17) | TrackBack

Response to Tsunami Comments--Posner

Let me respond briefly to a few comments that present interesting analytical issues.

One such issue is whether, given the large, perhaps infinite, number of low-probability disaster scanarios, it is possible to defend against all of them without bankrupting a nation, or the world. It is not possible to defend against all of them; indeed, it is not even possible to imagine all of them. Ideally, they should be ranked by expected cost (probability times loss), and the expected costs compared with the costs of prevention. A disaster-protection budget could then be determined and allocated across the different disasters in such a way as to minimize the total net expected costs. The approach will not work perfectly because of uncertainty about both probability and loss with respect to many of the possible disasters (and also means and costs of prevention, in many cases). But there appears to be no better approach.

My book Catastrophe: Risk and Response lists a number of disaster scenarios, with some effort at estimating probabilities, losses, and precaution costs. However, to create a comprehensive ranking along the lines suggested above would require consideration of a number of additional disaster scenarios--including tsunamis, which I mentioned only in passing in the book. Such a ranking would be a worthwhile research project.

A related point is that poor countries may not have the resources to create tsunami warning systems or take other precautionary measures that wealthy countries could afford to do. I would rephrase the point as follows: the budget for disaster prevention will depend on the competing claims on public and private funds. The more urgent those competing claims, the rationally smaller will be the budget devoted to disaster prevention. It is really the same point that I made in my original posting in noting that value of life estimates are positively correlated with per capita income. This is not because the lives of poor people are worth less in some ethical sense than those of the rich, or even that poor people consider that their life is worth less. One must understand that the value of life estimates, while useful in cost-benefit analysis, are really just arithmetic transformations of estimates of willingness to pay to avoid a risk of death. If a person will pay $70 to avoid a .00001 risk of death, we divide the first number by the second and call the resulting figure of $7 million the "value of life." The reason for the transformation is not to make an ethical point but to facilitate comparison between the costs and benefits of precautions.

Obviously a person who can avoid starvation only by taking a risky job will demand less to assume the risk than a rich person would. That is rational behavior and if we forbid him to take the risk and force him to starve we won't be doing him any favor.

But to keep matters in perspective, although per capita incomes in the nations affected by the recent Indian Ocean tsunami are roughly 10 times less than the per capita income of the United States, the four countries principally affected--Indonesia, Thailand, India, and Sri Lanka, have an aggregate GDP of hundreds of billions of dollars. A tsunami warning system might cost only a few million dollars a year.

Posted by posner at 05:43 PM | Comments (4) | TrackBack

January 05, 2005

The Tsunami and the Economics of Catastrophic Risk

The Indian Ocean tsunami illustrates a type of disaster to which policymakers pay too little attention—a disaster that has a very low or unknown probability of occurring, but that if it does occur creates enormous losses. Great as the death toll, physical and emotional suffering of survivors, and property damage caused by the recent tsunami are, even greater losses could be inflicted by other disasters of low (but not negligible) or unknown probability. The asteroid that exploded above Siberia in 1908 with the force of a hydrogen bomb might have killed millions of people had it exploded above a major city. Yet that asteroid was only about 200 feet in diameter, and a much larger one (among the thousands of dangerously large asteroids in orbits that intersect the earth’s orbit) could strike the earth and cause the total extinction of the human race through a combination of shock waves, fire, tsunamis, and blockage of sunlight, wherever it struck. Other catastrophic risks include, besides earthquakes such as the one that caused the recent tsunami, natural epidemics (the 1918–1919 Spanish influenza epidemic killed between 20 and 40 million people), nuclear or biological attacks by terrorists, certain types of lab accident, and abrupt global warming. The probability of catastrophes resulting, whether or not intentionally, from human activity appears to be increasing because of the rapidity and direction of technological advances.

The fact that a catastrophe is very unlikely to occur is not a rational justification for ignoring the risk of its occurrence. Suppose that a tsunami as destructive as the Indian Ocean one occurs on average once a century and kills 150,000 people. That is an average of 1,500 deaths per year. Even without attempting a sophisticated estimate of the value of life to the people exposed to the risk, one can say with some confidence that if an annual death toll of 1,500 could be substantially reduced at moderate cost, the investment would be worthwhile. A combination of educating the residents of low-lying coastal areas about the warning signs of a tsunami (tremors and a sudden recession in the ocean), establishing a warning system involving emergency broadcasts, telephoned warnings, and air-raid-type sirens, and improving emergency response systems, would have saved many of the people killed by the Indian Ocean tsunami, probably at a total cost below any reasonable estimate of the average losses that can be expected from tsunamis. Relocating people away from coasts would be even more efficacious, but except in the most vulnerable areas or in areas in which residential or commercial uses have only marginal value, the costs would probably exceed the benefits. For annual costs of protection must be matched with annual, not total, expected costs of tsunamis.

In my book Catastrophe: Risk and Response (Oxford University Press 2004), I try to be more precise about how one might determine the costs of catastrophes. There is now a substantial economic literature inferring the value of life from the costs people are willing to incur to avoid small risks of death; if from behavior toward risk one infers that a person would pay $70 to avoid a 1 in 100,000 risk of death, his value of life would be estimated at $7 million ($70/.00001), which is in fact the median estimate of the value of life of an American. Because value of life is positively correlated with income, this figure cannot be used to estimate the value of life of most of the people killed by the Indian Ocean tsunami. A further complication is that the studies may not be robust with respect to risks of death much smaller than the 1 in 10,000 to 1 in 100,000 range of most of the studies; we do not know what the risk of death from a tsunami was to the people killed. Additional complications come from the fact that undoubtedly more than 150,000 people have died or will die—and the total may never be known—and that there is vast suffering and property damage that must also be quantified, as well as estimates needed of just how effective precautionary measures of various scope and expense would have been. The risks of smaller but also still destructive tsunamis that such measures might protect against must also be factored in; nor am I confident about my “once a century” risk estimate. Nevertheless, it seems apparent that the total cost figure of the recent tsunami will come in at an amount great enough to indicate that there were indeed precautionary measures to take that would have been cost-justified.

Why, then, weren’t such measures taken in anticipation of a tsunami on the scale that occurred? Tsunamis are a common consequence of earthquakes, which themselves are common; and tsunamis can have other causes besides earthquakes—a major asteroid strike in an ocean would create a tsunami that would dwarf the Indian Ocean one.

There are a number of reasons for such neglect. First, although a once-in-a-century event is as likely to occur at the beginning of the century as at any other time, it is much less likely to occur in the first decade of the century than later. Politicians with limited terms of office and thus foreshortened political horizons are likely to discount low-risk disaster possibilities, since the risk of damage to their careers from failing to take precautionary measures is truncated. Second, to the extent that effective precautions require governmental action, the fact that government is a centralized system of control makes it difficult for officials to respond to the full spectrum of possible risks against which cost-justified measures might be taken. The officials, given the variety of matters to which they must attend, are likely to have a high threshold of attention below which risks are simply ignored. Third, where risks are regional or global rather than local, many national governments, especially in the poorer and smaller countries, may drag their heels in the hope of taking a free ride on the larger and richer countries. Knowing this, the latter countries may be reluctant to take precautionary measures and by doing so reward and thus encourage free riding. Fourth, countries are poor often because of weak, inefficient, or corrupt government, characteristics that may disable poor nations from taking cost-justified precautions. Fifth, people have difficulty thinking in terms of probabilities, especially very low probabilities, which they tend therefore to write off. This weakens political support for incurring the costs of taking precautionary measures against low-probability disasters.

The operation of some of these factors is illustrated by the refusal of the Pacific nations, which do have a tsunami warning system, to extend their system to the Indian Ocean prior to the recent catastrophe. Tsunamis are more common in the Pacific, and most of the Pacific nations do not abut on the Indian Ocean, but even if the risk of an Indian Ocean tsunami was only a tenth of that of a Pacific Ocean tsunami (a figure I have seen in a newspaper article), it was still worth taking precautions against; but there is a tendency to write down slight risks to zero.

An even more dramatic example of neglect of low-probability/high-cost risks concerns the asteroid menace, which is analytically similar to the menace of tsunamis. NASA, with an annual budget of more than $10 billion, spends only $4 million a year on mapping dangerously close large asteroids, and at that rate may not complete the task for another decade, even though such mapping is the key to an asteroid defense because it may give us years of warning. Deflecting an asteroid from its orbit when it is still millions of miles from the earth is a feasible undertaking. In both cases, slight risks of terrible disasters are largely ignored essentially for political reasons.

In part because tsunamis are one of the risks of an asteroid collision, the Indian Ocean disaster has stimulated new intereset in asteroid defense. This is welcome. The fact that a disaster of a particular type has not occurred recently or even within human memory (or even ever) is a bad reason to ignore it. The risk may be slight, but if the consequences should it materialize are great enough, the expected cost of disaster may be sufficient to warrant defensive measures.

Posted by posner at 01:04 AM | Comments (39) | TrackBack

Economic Effects of Tsunamis and Other Catastrophes

Economic Effects of Tsunamis and Other Catastrophes- Becker
A Reaction to Posner's Discussion

John Stuart Mill, the great 19TH century English economist and philosopher, optimistically, but I believe accurately, remarked on “…the great rapidity with which countries recover from a state of devastation, the disappearance in a short time, of all traces of mischief done by earthquakes, floods, hurricanes, and the ravages of war”. The history of both natural and man-made disasters during the subsequent century and a half generally supports Mill.

The 9/11 terrorist attacks were quickly recognized as the beginning of a series of possibly more destructive attacks on US citizens and property, and many commentators then believed it would cause a major economic depression. Yet it had a slight overall impact on the course of GDP and employment in the United States, although some industries and New York City were affected for several years. The Kobe earthquake of 1995 killed over 6000 persons, and destroyed more than 100,000 homes, still the economic recovery not only of Japan but also of the Kobe economy was rapid. The flu pandemic of 1918-19 killed about 30 million persons worldwide without having a major impact on the world’s economy. The lasting economic effects are similarly small for most other natural disasters that have occurred during the past couple of centuries.

Many natural catastrophes have very low probabilities of occurring, but cause considerable destruction of both life and property when they do happen. The recent tsunami in the Indian Ocean is one horrible example: it killed many more people than either 9/11 or the Kobe earthquake. But bad as it is, the loss of life is much smaller relative to the populations of the nations affected than some previous disasters. For example, the Lisbon earthquake of 1755 may have killed 60,000 people, other earthquakes in the past are alleged to have killed in the hundreds of thousands, and I mentioned the flu epidemic of 1918-19 that killed tens of millions worldwide.

History and analysis both indicate that the economic recovery of the nations most adversely affected by this tsunami will be rapid, although it will take longer in the resorts and coastal regions hit the hardest. The expectation of rapid recovery explains why Asian stock markets did not change much after the tsunami struck: Indonesia’s and Malaysia’s actually rose a little during the last week of December, while Thailand’s declined a little, and Sri Lanka’s declined by a few per cent.

I fully agree with Posner that it is worth spending considerably more to provide better early warning systems about the future occurrence of earthquakes and tsunamis, asteroids that might strike the earth, and other catastrophes. But no matter how much is spent and how much planning takes place, natural catastrophes will continue and will sometimes be unexpected.

There are two ways to protect against natural and other disasters: one is through insurance that helps compensate persons badly hurt by loss of family member or property. The other is through self-protection, which means actions to reduce the probability of the disasters from happening- as when a person drives more carefully to reduce the likelihood of getting into an accident, or when countries agree to reduce emissions of greenhouse gases in the hope of reducing the probability of severe global warming.

As more is learnt about various natural disasters, more self-protective actions would become available. But for many of the very infrequent ones, even given generous estimates of the value of life of the type discussed by Posner, it does not pay to take expensive self-protection actions. The best response in these cases is to have an effective insurance system for those badly harmed. So I concentrate most of my comments about protection against disasters on insurance.

Survivors of disasters that strike rich nations usually have medical coverage to pay for their treatment and rehabilitation, and insurance to cover much of their property destroyed, while those who perish usually leave life insurance for their families, and the opportunity to obtain decent education for their children. By contrast, most individuals in poor nations of Asia and elsewhere mainly rely on help from their families and neighbors when disasters strike. Unfortunately, such help is not available when disasters attack many members of the same family and whole neighborhoods, as in major tsunamis and earthquakes.

An effective way for poorer nations to respond in the longer run would be to encourage greater investment in education. Since education raises the earnings of individuals and the per capita incomes of countries, education clearly makes it easier to cope with disasters- as Mill had already recognized when he emphasizes the importance of knowledge in hastening the recovery from disasters. Beyond that, however, my colleague at the University of Chicago, Casey Mulligan, and I have shown that educated persons take a much longer time perspective in their personal decisions. This means that they are much more likely to anticipate the incidence and location of natural catastrophes when they decide where to live and how their houses are built, and they better self-protect and self-insure themselves in other ways as well.

But in the short run, greater access to private market insurance, even if subsidized by governments, is important. Regrettably, such insurance is not likely to be available to, or chosen, by the type of very poor families disproportionately affected by this tsunami. The large outpouring of aid from rich nations will help only temporarily in the very near term. The next best alternative to private insurance would be government disaster programs in poor countries that designate areas hit by major earthquakes, hurricanes, and other catastrophes, man-made or natural, as eligible for disaster assistance. Such programs could make sufficient payments to poor families of husbands and fathers who died, and to families that lost most of their property, to help put them on their economic feet, without causing much of a drain on the government budgets of even poorer developing nations like Indonesia and Sri Lanka.

The moral hazard effects of such programs are always worrisome- families might continue to build homes on earthquake fault lines if they expect government compensation when their homes are destroyed, or continue to build close to the shore in potential water-borne disaster areas. There is no perfect offset against such rational responses to government coverage of losses, but incomplete protection (“co-payments”), and regulatory exclusion of certain types of construction and other vulnerable activities in potential disaster regions would encourage individuals to consider the risks involved in their actions.

Posted by becker at 12:49 AM | Comments (8) | TrackBack

January 04, 2005

Tsunami

We will be posting early Wednesday morning slightly fuller versions of the op-ed pieces on the Indian Ocean tsunami that we published in the Wall Street Journal today.

Posted by posner at 08:47 PM | Comments (0) | TrackBack