Marketplace of Perceptions

50 March - April 2006

Like all revolutions in thought, thisone began with anomalies, strange facts,odd observations that the prevailing wis-dom could not explain. Casino gamblers,for instance, are willing to keep bettingeven while expecting to lose. People saythey want to save for retirement, eat better,start exercising, quit smokingand theymean itbut they do no such things. Vic-tims who feel theyve been treated poorlyexact their revenge, though doing so hurtstheir own interests.

Such perverse facts are a direct aront tothe standard model of the human actorEconomic Manthat classical and neo-classical economics have used as a founda-tion for decades, if not centuries. EconomicMan makes logical, rational, self-interesteddecisions that weigh costs against benetsand maximize value and prot to himself.Economic Man is an intelligent, analytic, selsh creature who hasperfect self-regulation in pursuit of his future goals and isunswayed by bodily states and feelings. And Economic Man is amarvelously convenient pawn for building academic theories. ButEconomic Man has one fatal aw: he does not exist.

When we turn to actual human beings, we nd, instead ofrobot-like logic, all manner of irrational, self-sabotaging, and even

altruistic behavior. This is such a routine ob-servation that it has been made for cen-turies; indeed, Adam Smith saw psychologyas a part of decision-making, says assistantprofessor of business administration NavaAshraf. He saw a conict between the pas-sions and the impartial spectator.

Nonetheless, neoclassical economicssidelined such psychological insights. Asrecently as 15 years ago, the sub-disciplinecalled behavioral economicsthe studyof how real people actually make choices,which draws on insights from both psy-chology and economicswas a marginal,exotic endeavor. Today, behavioral eco-nomics is a young, robust, burgeoningsector in mainstream economics, and canclaim a Nobel Prize, a critical mass of em-pirical research, and a history of upendingthe neoclassical theories that dominated

the discipline for so long.Although behavioral economists teach at Stanford, Berkeley,

Chicago, Princeton, MIT, and elsewhere, the subelds greatestconcentration of scholars is at Harvard. Harvards approach toeconomics has traditionally been somewhat more worldly andempirical than that of other universities, says PresidentLawrence H. Summers, who earned his own economics doctor-

The Marketplace of

PerceptionsBehavioral

economics explainswhy we

procrastinate, buy,borrow, and

grab chocolate on the spur of the

moment.

by Craig Lambert

Portraits by Stu Rosner

Beh.Eco-final 2/10/06 8:59 AM Page 50

Harvard Magazine 51

ate at Harvard and identies himselfas a behavioral economist. And ifyou are worldly and empirical, youare drawn to behavioral approaches.

Framing a New FieldTwo non-economists have wonNobel Prizes in economics. As earlyas the 1940s, Herbert Simon ofCarnegie Mellon University put for-ward the concept of bounded ratio-nality, arguing that rational thoughtalone did not explain human deci-sion-making. Traditional economistsdisliked or ignored Simons research,and when he won the Nobel in 1978,many in the field were very unhappyabout it.

Then, in 1979, psychologists DanielKahneman, LL.D. 04, of Princetonand Amos Tversky of Stanford pub-lished Prospect Theory: An Analysisof Decision under Risk, a break-through paper on how people handleuncertain rewards and risks. In theensuing decades, it became one of themost widely cited papers in econom-ics. The authors argued that the waysin which alternatives are framednotsimply their relative valueheavilyinuence the decisions people make.This was a seminal paper in behav-ioral economics; its rigorous equa-tions pierced a core assumption ofthe standard modelthat the actualvalue of alternatives was all that mat-tered, not the mode of their presenta-tion (framing).

Framing alternatives dierentlycan, for example, change peoplespreferences regarding risk. In a 1981Science paper, The Framing of Deci-sions and the Psychology of Choice,Tversky and Kahneman presented anexample. Imagine that the U.S. ispreparing for the outbreak of an un-usual Asian disease which is ex-pected to kil l 600 people, theywrote. Two alternative programs tocombat the disease have been pro-posed. Choose Program A, and aprojected 200 people will be saved.Choose Program B, and there is aone-third probability that 600 peoplewill be saved, and a two-thirds probability that no one will besaved. The authors reported that 72 percent of respondents choseProgram A, although the actual outcomes of the two programsare identical. Most subjects were risk averse, preferring the cer-

tain saving of 200 lives. The researchers then restated the prob-lem: this time, with Program C, 400 people will die, whereaswith Program D, there is a one-third probability that no onewill die, and a two-thirds probability that 600 people will die.

Professor of economics David Laibson,whose research explores the fundamentaltension between seizing available rewardsin the present, and being patient for rewards in the future



This time, 78 percent chose Program Dagain, despite identicaloutcomes. Respondents now preferred the risk-taking option.The dierence was simply that the rst problem phrased its op-tions in terms of lives saved, and the second one as lives lost; peo-ple are more willing, apparently, to take risks to prevent livesbeing lost than to save lives.

Kahneman and Tversky started this revolution in econom-ics, says Straus professor of business administration Max Bazer-man, who studies decision-making and negotiation at HarvardBusiness School. That 1979 paper was written on the turf of eco-nomics, in the style of economists, and published in the toughesteconomic journal, Econometrica. The major points of prospect the-

ory arent hard to state in words. The math was added for accep-tance, and that was important. In 2002, Kahneman received theNobel Prize in economics along with Vernon Smith, Ph.D. 55, ofGeorge Mason University, who was honored for work in experi-mental economics. (Tversky, Kahnemans longtime collaborator,had died in 1996.)

In the 1980s, Richard Thaler (then at Cornell, now of the Uni-versity of Chicago Graduate School of Business) began import-ing such psychological insights into economics, writing a regu-lar feature called Anomalies in the Journal of Economic Perspectives(later collected in his 1994 book, The Winners Curse). Dick Thalerlived in an intellectual wilderness in the 1980s, says professor ofeconomics David Laibson, one of Har-vards most prominent behavioraleconomists. He championed theseideas that economists were deriding.But he stuck to it. Behavioral ap-proaches were anathema in the 1980s,became popular in the 1990s, andnow were a fad, with lots of gradstudents coming on board. Its nolonger an isolated band of belea-guered researchers ghting againstthe mainstream.

As with most movements, therewere early adopters. In the 1980s thebest economists in the world wereseeing the evidence and adopting it[behavioral economics], Bazermansays. Mediocre economists followslowlythey continued to ignore it sothey could continue doing their workundisturbed.

To be fair, the naysayers would haveagreed that the rational model only ap-proximates human cognitionjust asNewtonian physics is an approxima-tion to Einsteins physics, Laibson ex-

plains. Although there are dierences, when walking along thesurface of this planet, youll never encounter them. If I want tobuild a bridge, pass a car, or hit a baseball, Newtonian physicswill suce. But the psychologists said, No, its not sucient,were not just playing around at the margins, making smallchange. There are big behavioral regularities that include thingslike imperfect self-control and social preferences, as opposed topure selshness. We care about people outside our families andgive up resources to help themthose aected by Hurricane Ka-trina, for example.

Much of the early work in behavioral economics was in nance,with many signicant papers written by Jones professor of eco-

nomics Andrei Shleifer. In nancialmarkets, The usual arguments inconventional economics are, This[behavioral irrationality] cant betrue, because even if there are stu-pid, irrational people around, theyare met in the marketplace by smart,rational people, and trading by thesearbitrageurs corrects prices to ratio-

nal levels, Shleifer explains. For example, if people get undulypessimistic about General Motors and dump GM shares on themarket, these smart people will sweep in and buy them up as un-dervalued, and not much will happen to the price of GM shares.

But a 1990 paper Shleifer wrote with Summers, The NoiseTrader Approach to Finance, argues against this ecient mar-ket model by noting that certain risk-related factors limit thisarbitrage. At that time, for example, shares of Royal Dutch wereselling at a dierent price in Amsterdam than shares of Shell inLondon, even though they were shares of the same company,Royal Dutch/Shell. Closed-end mutual funds (those with a xednumber of shares that trade on exchanges) sell at dierent prices

than the value of their portfolios. Whenthe same thing sells at two dierentprices in dierent markets, forces of ar-bitrage and rationality are necessarilylimited, Shleifer says. The forces of ir-rationality are likely to have a big im-pact on prices, even on a long-termbasis. This is a theoretical attack onthe central conventional premise.

Meanwhile, the Russell Sage Foun-dation, which devotes itself to researchin the social sciences, consistently sup-ported behavioral economics, evenwhen it was in the intellectual wilder-ness. Current Sage president Eric Wan-ner, Ph.D. 69, whose doctorate is in so-cial psychology, was running a programin cognitive science at the Alfred P.Sloan Foundation in 1984 when Sloanstarted a behavioral economics programas an application of cognitive science tothe study of economic decision-making.(The eld is misnamedit should havebeen called cognitive economics, saysWanner. We werent brave enough.)After Wanner became president of Rus-

Now, we want chocolate, cigarettes, and a trashy movie. In the future, we want to eat fruit, to quit smoking, and to watch Bergman films.

D a v i d L a i b s o n

ST

U R

OS

NE

R


Harvard Magazine 53

sell Sage in 1986, the two institutions worked jointly to fosterthe new subeld. In the last 20 years, Sage has made well over100 grants to behavioral economists; it also organizes a biennialsummer institute that has drawn younger scholars like Laibsonand professor of economics Sendhil Mullainathan. PrincetonUniversity Press and Russell Sage also co-publish a series ofbooks in the eld.

Behavioral economics, then, is the hybrid ospring of economicsand psychology. We dont have much to tell psychologists abouthow individuals make decisions or process information, but wehave a lot to learn from them, says Glimp professor of economicsEdward Glaeser. We do have a lot to say about how individualscome together in aggregationsmarkets, rms, political parties.

The Seductive Now-MomentA national chain of hamburger restaurants takes its namefrom Wimpy, Popeyes portly friend with a voracious appetitebut small exchequer, who made famous the line, Ill gladly payyou Tuesday for a hamburger today. Wimpy nicely exempliesthe problems of intertemporal choice that intrigue behavioral

economists like David Laibson. Theres a fundamental tension,in humans and other animals, between seizing available rewardsin the present, and being patient for rewards in the future, hesays. Its radically important. People very robustly want instantgratication right now, and want to be patient in the future. Ifyou ask people, Which do you want right now, fruit or choco-late? they say, Chocolate! But if you ask, Which one a weekfrom now? they will say, Fruit. Now we want chocolate, ciga-rettes, and a trashy movie. In the future, we want to eat fruit, toquit smoking, and to watch Bergman lms.

Laibson can sketch a formal model that describes this dy-namic. Consider a project like starting an exercise program,which entails, say, an immediate cost of six units of value, butwill produce a delayed benet of eight units. Thats a net gain oftwo units, but it ignores the human tendency to devalue the fu-ture, Laibson says. If future events have perhaps half the value ofpresent ones, then the eight units become only four, and startingan exercise program today means a net loss of two units (sixminus four). So we dont want to start exercising today. On theother hand, starting tomorrow devalues both the cost and the

Assistant professor of business administration Nava Ashraf helpedadapt a home-grown savings techniqueshe saw in West Africa to the Philippines, where the cute SEED(Save, Earn, Enjoy Deposits) bank (opposite) helped ordinary citizens save money.



benet by half (to three and four units, respectively), resulting ina net gain of one unit from exercising. Hence, everyone is enthusi-astic about going to the gym tomorrow.

Broadly speaking, People act irrationally in that they overlydiscount the future, says Bazerman. We do worse in life be-cause we spend too much for what we want now at the expenseof goodies we want in the future. People buy things they cantaord on a credit card, and as a result they get to buy less overthe course of their lifetimes. Such problems should not arise, ac-cording to standard economic theory, which holds that thereshouldnt be any disconnect between what Im doing and what Iwant to be doing, says Nava Ashraf.

Luckily, Odysseus also confronts the problem posed byWimpyand Homers hero solves the dilemma. The goddessCirce informs Odysseus that his ship will pass the island of theSirens, whose irresistible singing can lure sailors to steer towardthem and onto rocks. The Sirens are a marvelous metaphor forhuman appetite, both in its seductions and its pitfalls. Circe ad-vises Odysseus to prepare for temptations to come: he must orderhis crew to stopper their ears with wax, so they cannot hear the

Sirens songs, but he may hear the Sirens beautiful voices withoutrisk if he has his sailors lash him to a mast, and commands themto ignore his pleas for release until they have passed beyond dan-ger. Odysseus pre-commits himself by doing this, Laibson ex-plains. Binding himself to the mast prevents his future self fromcountermanding the decision made by his present self.

Pre-commitments of this sort are one way of getting aroundnot only the lure of temptation, but our tendency to procrasti-nate on matters that have an immediate cost but a future payo,like dieting, exercise, and cleaning your oce. Take 401(k) retire-ment plans, which not only let workers save and invest for retire-ment on a tax-deferred basis, but in many cases amount to a bo-nanza of free money: the equivalent of nding $100 Bills on theSidewalk (the title of one of Laibsons papers, with James Choiand Brigitte Madrian). Thats because many rms will match employees contributions to such plans, so one dollar becomestwo dollars. Its a lot of free money, says Laibson, who has pub-lished many papers on 401(k)s and may be the worlds foremostauthority on enrollment in such plans. Someone making$50,000 a year who has a company that matches up to 6 percent

Certain patterns of response torewards seem to be biologically embedded in the human brain. Abranch of behavioral economicscalled neuroeconomics looks inside thebrain with scanning tools like mag-netic resonance imaging (MRI) to investigate patterns of mo-tivation. Neuroeconomics is controversial, as the link betweencerebral blood ow and decision-making is less straightfor-ward than, say, playing slot machines and losing money. Yet itis one of the most fascinating and provocative aspects of theyoung eld.

Economists specialize in taking really complex things andboiling them down to simple principles, says David Laibson.So, rather than treat the brain as billions of neurons, or trillionsof neurotransmitters, we want to ask, what is the right level ofanalysis? It turns out that the brain has two key subsystems.One, the limbic and paralimbic system, rules theintuitive and aective parts of our psyches.Its shared by all mammals and seems todo a lot of emotional cognitionhowwe feel emotionally, how we re-spond to other humans, or tobeing treated unfairly. This sys-tem seems to function uncon-sciously; we dont have accessto it and maybe cant even con-trol it. Its experiential andrapid in function.

Contrast that with the ana-lytic system, centered in thefrontal and parietal cortexes, Laib-son continues. It controls a lot of thethought processes we learn to do: cal-culated, conscious, future-oriented

thinking. Its not based on past expe-rience; you could have the rules of abrand-new game explained and theanalytic system would be able togure out how to play.

Brain researchers have shown thatan interaction of the limbic and analytic systems governshuman decision-making. The limbic system seems to radicallydiscount the future. While the analytic systems role remainsconstant from the present moment onward, the limbic systemassumes overriding importance in the present moment, butrapidly recedes as rewards move into the future and the emo-tional brain reduces its activation. This explains impulsive-ness: the slice of pizza thats available right now trumps the di-etary plan that the analytic brain has formulated. Seizingavailable rewards now might be a response pattern with evo-lutionary advantages, as future benets are always uncertain.

Consider an experiment that scans thebrains of research subjects oered a

choice between present and future re-wards: $20 now, or $23 a month from

now. Both limbic and analytic sys-tems show activity. Then changethe oer to two future prospects:$20 two weeks from now, or $23 ina month. In this case, the limbicsystem pretty much drops out.The analytic system, in contrast,shows the same activation pat-

terns regardless of the delay, be ithours or months. When the ana-

lytic system is more active, peoplechoose the patient reward; whenboth systems are active, temptation

usually trumps prudence.

Neuroeconomics

GE

TT

Y I

MA

GE

S


Harvard Magazine 55

of his contributions could receive an additional $3,000 per year.The rational model unequivocally predicts that people will cer-

tainly snap up such an opportunity. But they dontnot evenworkers aged 59 or older, who can withdraw sums from their401(k) plans without penalty. (Younger people are even more un-likely to contribute, but they face apenalty for early withdrawal.) Itturns out that about half of U.S.workers in this [above 59 ] agegroup, who have this good dealavailable, are not contributing,says Laibson. Theres no down-side and a huge upside. Still, indi-viduals are procrastinatingtheyplan to enroll soon, year after year,but dont do it. In a typical Ameri-can rm, it takes a new employee amedian time of two to three yearsto enroll. But because Americanschange jobs frequentlysay, everyve yearsthat delay could meanlosing half of ones career opportu-nity for these retirement savings.

Laibson has run educational in-terventions with employees atcompanies, walking them throughthe calculations, showing themwhat they are doing wrong. Al-most all of them still dont invest,Laibson says. People nd thesekinds of nancial transactions un-pleasant and confusing, and theyare happier with the idea of doingit tomorrow. It demonstrates howpoorly the standard rational-actormodel predicts behavior.

Its not that we are utterly help-less against procrastination. Laib-son worked with a rm thatforced its employees to make activedecisions about 401(k) plans, insist-ing on a yes or no answer within30 days. This is far dierent fromgiving people a toll-free phonenumber to call whenever they de-cide to enroll. During the 30-dayperiod, the company also sent fre-quent e-mail reminders, pressur-ing the sta to make their deci-sions. Under the active-decisionplan, enrollment jumped from 40to 70 percent. People want to beprudent, they just dont want todo it right now, Laibson says.Youve got to compel action. Orenroll people automatically.

When he was U.S. Treasury Sec-retary, Lawrence Summers appliedthis insight. We pushed very hard

for companies to choose opt-out [automatic enrollment] 401(k)srather than opt-in [self-enrollment] 401(k)s, he says. In classicaleconomics, it doesnt matter. But large amounts of empirical evi-dence show that defaults do matter, that people are inertial, andwhatever the baseline settings are, they tend to persist.

Associate professor of public policy Iris Bohnet, who has played games thatmeasure aversion to betrayal with subjects from Brazil to Switzerland to Kuwait



Marketing PrudenceThese insights can also be writ large. Laibsons former studentNava Ashraf, who has worked extensively with non-governmen-tal organizations, is now applying behavioral economics to inter-ventions in developing countries. She lived for a year in IvoryCoast and Cameroon, where she noticed that farmers andsmall-business owners were often not doing the things that a de-velopment policymaker or economist thinks they should do, shesays. They wouldnt take up technologies that would increaseagricultural yield, for example. They wouldnt get vaccines, eventhough they were free! They also had a lot of trouble saving. InJanuary they had a lot of money and would spend it on feasts andspecial clothes, but in June their children would be starving.

Still, some found ways to oset their less-than-prudent ten-dencies. One woman had a cashbox in her home, where shesaved money regularlyand gave her neighbor the only key.Another timed the planting of her sweet-potato crop so thatthe harvest would come in when school fees were due. Herfarm became an underground bank account that allowed with-drawal only at the proper moment.

Ashraf worked with a bank in the Philippines to design a sav-ings plan that took o from the African womans cashbox. The

bank created a savings account, called SEED (Save, Earn, EnjoyDeposits), with two features: a locked box (for which the bankhad the key) and a contractual agreement that clients could notwithdraw money before reaching a certain date or sum. Theclients determined the goal, but relied on the bank to enforce thecommitment. The bank marketed the SEED product to literateworkers and micro-entrepreneurs: teachers, taxi drivers, peoplewith pushcart businesses.

The SEED box, designed to appeal to the banks clients (Inthe Philippines, they like cute stuff, Ashraf explains ), helpedmobilize deposits. Its similar to automatic payroll deduction,but not enough of the customers had direct deposit to makethat work, she says. To further encourage deposits, Ashrafworked with the bank on an additional program of deposit col-lectors who, for a nominal fee, would go to the customers homeon a designated day and collect the savings from the SEED box.The withdrawal restrictions on the account helped clients avoidthe temptation of spending their savings. The SEED savings ac-count made a designed choice available in the marketplace that,so far, has helped a growing number of microfinance clients inthe Philippines reach their savings goals.

Ashraf is now working with Population Services Interna-

Policymakers think that if they get the abstractions right, that willdrive behavior in the desired direction, says professor of economicsSendhil Mullainathan. But the world happens in real time.


Harvard Magazine 57

tionala nonprot organization that seeks to focus private-sec-tor resources on the health problems of developing nationsona project in Zambia to motivate people to use a water puri-cation solution known as Clorin. We can use what marketingpeople have known all along, Ashraf says. There are ways ofmanipulating peoples psychological frameworks to get them tobuy things. How do you use this knowledge to get them to adoptsocially useful products or services? Its so practical, and very im-portant in development, for anybody who wants to help peoplereach their goals.

Carefully designed programs like the SEED bank are examplesof what Richard Thaler called prescriptive economics, whichaims not only to describe the world but to change it. Behavioraleconomics really shines when you talk about the specics ofwhat the policy should look like, says Sendhil Mullainathan, whoreceived a MacArthur Fellowshipin 2002. The dierence in impactbetween two broad policies maynot be as great as dierences inhow each policy is frameditsdeadlines, implementation, and thedesign of its physical appearance.

For example, in Social Security privatization, Mullainathancontinues, the dierence between private accounts and the sta-tus quo may be less than that between two dierent ways of im-plementing private accounts. What is the default option? Areyou allowed to make changes? Whats the deadline for makingchanges? How are the monthly statements presentedjust yourreturns, or are the market returns printed alongside your own?In terms of impact, the devil really is in the details of how theprogram is designed. We know that people have a tough timemaking these choices. So how are the choices framed? Whatmetrics do they focus on?

We tend to think people are driven by purposeful choices, heexplains. We think big things drive big behaviors: if people dontgo to school, we think they dont like school. Instead, most behav-iors are driven by the moment. They arent purposeful, thought-out choices. Thats an illusion we have about others. Policymak-ers think that if they get the abstractions right, that will drivebehavior in the desired direction. But the world happens in realtime. We can talk abstractions of risk and return, but when theperson is physically checking o the box on that investment form,all the things going on at that moment will disproportionatelyinuence the decision they make. Thats the temptation ele-mentin real time, the moment can be very tempting. The mainthing is to dene what is in your mind at the moment of choice.Suppose a company wants to sell more soap. Traditional econo-mists would advise things like making a soap that people likemore, or charging less for a bar of soap. A behavioral economistmight suggest convincing supermarkets to display your soap ateye levelpeople will see your brand rst and grab it.

Mullainathan worked with a bank in South Africa thatwanted to make more loans. A neoclassical economist wouldhave oered simple counsel: lower the interest rate, and peoplewill borrow more. Instead, the bank chose to investigate somecontextual factors in the process of making its oer. It mailedletters to 70,000 previous borrowers saying, Congratulations!Youre eligible for a special interest rate on a new loan. But theinterest rate was randomized on the letters: some got a low rate,

others a high one. It was done like a randomized clinical trial ofa drug, Mullainathan explains.

The bank also randomized several aspects of the letter. In onecorner there was a photovaried by gender and raceof a bankemployee. Dierent types of tables, some simple, others complex,showed examples of loans. Some letters oered a chance to win acell phone in a lottery if the customer came in to inquire about aloan. Some had deadlines. Randomizing these elements allowedMullainathan to evaluate the eect of psychological factors as op-posed to the things that economists care abouti.e., interestratesand to quantify their eect on response in basis points.

What we found stunned me, he says. We found that anyone of these things had an eect equal to one to ve percentagepoints of interest! A womans photo instead of a mans increaseddemand among men by as much as dropping the interest rate ve

points! These things are not small. And this is very much an eco-nomic problem. We are talking about big loans here; customerswould end up with monthly loan payments of around 10 percentof their annual income. Youd think that if you really needed themoney enough to pay this interest rate, youre not going to beaected by a photo. The photo, cell phone lottery, simple or com-plicated table, and deadline all had eects on loan applicationscomparable to interest. Interest rate may not even be the thirdmost important factor. As an economist, even when you thinkpsychology is important, you dont think its this important. Andchanging interest rates is expensive, but these psychological ele-ments cost nothing.

Mullainathan is helping design programs in developing coun-tries, doing things like getting farmers to adopt better feed forcows to increase their milk production by as much as 50 percent.Back in the United States, behavioral economics might be able toraise compliance rates of diabetes patients, who dont alwaystake prescribed drugs, he says. Poor families are often deterredfrom applying to colleges for nancial aid because the forms aretoo complicated. An economist would say, With $50,000 atstake, the forms cant be the obstacle, he says. But they can. (Atraditional explanation would say that the payo clearly out-weighs the cost in time and eort, so people wont be deterred bycomplex forms.)

Economists and others who engage in policy debates like towrangle about big issues on the macroscopic level. The nitty-grittydetails of executionwhat do the forms look like? what is in thebrochures? how is it communicated?are left to the support sta.But that work is central, Mullainathan explains. There shouldbe as much intellectual energy devoted to these design choices asto the choice of a policy in the rst place. Behavioral economicscan help us design these choices in sensible ways. This is a big holethat needs to be lled, both in policy and in science.

The Supply of HatredWhile some try to surmount or cope with irrationality, oth-ers feed upon it. In the wake of

An economist would say, With $50,000 at stake, the forms cant be the obstacle.But they can.

(please turn to page 93)

S e n d h i l M u l l a i n a t h a n


Harvard Magazine 93

the 9/11 attacks, Edward Glaeser began using behavioral eco-nomic approaches to research the causes of group hatred thatcould motivate murderous acts of that type. An economistsdenition of hatred, he says, is the willingness to pay a priceto inict harm on others. In laboratory settings, social scien-tists have observed subjects playing the ultimatum game, inwhich, say, with a total kitty of $10, Player A oers to split thecash with player B. If B accepts As oer, they divide the moneyaccordingly, but if B rejects As oer, both players get nothing.In thousands of trials around the world, with dierent stakes,people reject oers of 30 percent [$3 in our example] or less,says Glaeser. So typically, people oer 40 or 50 percent. But aconventional economic model would say that B should accept asplit of even one cent versus $9.99, since you are still better owith a penny than nothing. (If a computer, rather than ahuman, does the initial split, player B is much more likely to ac-cept an unfair splita conrmation of research conducted byprofessors at the Kennedy School of Government; see Games ofTrust and Betrayal, page 94.)

Clearly, the B player is willing to suer nancial loss in orderto take revenge on an A player who is acting unfairly. You dontpoke around in the dark recesses of human behavior and not ndvengeance, Glaeser says. Its pretty hard to nd a case of mur-der and not nd vengeance at the root of it.

The psychological literature, he found, denes hatred as anemotional response we have to threats to our survival or repro-duction. Its related to the belief that the object of hatred hasbeen guilty of atrocities in the past and will be guilty of themin the future, he says. Economists have nothing to tell psy-chologists about why individuals hate. But group-level hatredhas its own logic that always involves stories about atrocities.These stories are frequently false. As [Nazi propagandistJoseph] Goebbels said, hatred requires repetition, not truth, tobe eective.

You have to investigate the supply of hatred, Glaeser contin-ues. Who has the incentive and the ability to induce group ha-tred? This pushes us toward the crux of the model: politicians oranyone else will supply hatred when hatred is a complement totheir policies. Glaeser searched back issues of the Atlanta Consti-tution from 1875 to 1925, counting stories that contained the key-words Negro + rape or Negro + murder. He found a time-se-ries that closely matched that for lynchings described byhistorian C. Vann Woodward: rising from 1875 until 1890, reach-ing a plateau from 1890 until 1910, then declining after 1910.

In the 1880s and 1890s, Glaeser explains, the southern Populist

Party favored large-scale redistribution of wealth from the rich tothe poor, and got substantial support from African Americans.Wealthier Southern conservatives struck back, using race hatredand spreading untrue stories about atrocities perpetrated byblacks, Glaeser says. Populists are friends of blacks, and blacksare dangerous and hateful, was the messageinstead of beingsupported, [blacks] should be sequestered and have their re-sources reduced. [Rich whites] sold this to poor white voters,winning votes and elections. Eventually the Populists gave in anddecided they were better o switching their appeal to poor, racistwhites. They felt it was better to switch policies than try tochange voters opinions. The storiesall about rape and murderwere coming from suppliers who were external to poor whites.

Glaeser applies this model to anti-American hatred, which, indegree, is not particularly correlated with places that the UnitedStates has helped or done harm to, he says. France hates Americamore than Vietnam does. Instead, he explains, it has much to dowith political entrepreneurs who spread stories about past andfuture American crimes. Some place may have a leader who has aworking relationship with the United States. Enemies of the leaderoer an alternative policy: completely break with the United Statesand Israel, and attack them. We saw it in the religious enemies ofthe shah [of Iran]. The ayatollah sought to discredit the secularmodernists through the use of anti-American hatred.

For Glaeser, behavioral economics can take something we havefrom psychologyhatred as a hormonal response to threatsandput this in a market setting. What are the incentives that will in-

crease the supply of hatred in aspecic setting? Economists, hefeels, can take human tendenciesrooted in hormones, evolution, andthe stable features of social psychol-ogy, and analyze how they will playout in large collectivities. Much ofpsychology shows the enormoussensitivity of humans to socialinuence, Glaeser says. The Mil-gram and Zimbardo experiments

[on obedience to authority and adaptation to the role of prisonguard] show that humans can behave brutally. But that doesnt ex-plain why Nazism happened in Germany and not England.

Zero-Sum PersuasionAndrei shleifer has already made path-breaking contribu-tions to the literatures of behavioral nance (as noted above),political economy, and law and economics. His latest obsession ispersuasionHow people absorb information and how they aremanipulated, he says. At the American Economic Associationmeetings in January, Shleifer described cognitive persuasion,exploring how advertisers, politicians, and others attach theirmessages to pre-existing maps of associations in order to movethe public in a desired direction.

The Marlboro Man, for example, sold ltered cigarettes bymobilizing the publics associations of cowboys and the Westwith masculinity, independence, and the great outdoors. Thereis a conrmation bias, Shleifer explained, which favors persua-sive messages that conrm beliefs and connections already in theaudiences mind (see The Market for News, January-February,page 11, on work by Shleifer and Mullainathan that applies a sim-

You have to investigate the supply of hatred. Who has the incentive and ability to induce group hatred? Politicians or anyone else will supply hatred when hatred complements their policies.

THE MARKETPLACE OF PERCEPTIONS(continued from page 57)

E d w a r d G l a e s e r



ilar analysis to the news media). For example, George W. Bushwearing a $3,000 cowboy hat was not a problem, because itmatched his image, but John Kerry riding a $6,000 bicycle was aproblemthat luxury item appeared hypocritical for a candi-date claiming to side with the downtrodden.

Citing Republican pollster and communications consultantFrank Luntz, Shleifer noted how the estate tax was renamed thedeath tax (although there is no tax on death) in order to suc-cessfully sell its repeal. The relabeling linked the tax to the un-pleasant associations of the word death, and the campaignasked questions like, How can you burden people even more atthis most dicult time in their lives? Messages, not hard at-tributes, shape competition, Shleifer said; he noted that the fearof terrorism is a bigger issue in probable non-target states likeWyoming, Utah, and Nevada than in New York and New Jersey.

Because successful persuasive messages are consistent withprevailing worldviews, one corollary of Shleifers analysis is thatpersuasion is denitely not education, which involves addingnew information or correcting previous perceptions. Dont tellpeople, You are stupid, and here is what to think, Shleifersaid. During presidential debates, he asserted, voters tune out orforget things that are inconsistent with their beliefs. Educa-tional messages may be doomed, he added. They do not res-onate. In economic and political markets, he said, there is notendency toward a median taste; divergence, not convergence, isthe trend. Therefore, the successful persuader will nd a nicheand pander to it.

When making choices in the marketplace, People are not re-sponding to the actual objects they are choosing between, says EricWanner of the Russell Sage Foundation. There is no direct rela-

People care not only about outcomes,but about how outcomes came to be, says as-sociate professor of public policy Iris Bohnetof the Kennedy School of Government. Thatdoesnt strike anyone but an economistlikemeas a surprise. Game theory, as concep-tualized by conventional economics, suggeststhat players care only about substantive re-sults. With Ramsey professor of politicaleconomy Richard Zeckhauser, Bohnet devel-oped a concept of betrayal aversion, build-ing on the well-established psychologicalprinciple of risk aversionby and large, hu-mans simply dont like to take risks.

It turns out they dont like to trust, either, because trust is aform of risk that makes one vulnerable to betrayal. To buy anitem on eBay, one must trust the seller. We also trust attorneys,doctors, and politicians to tell us the truth and to represent ourinterests. These are principal-agent relationships, Bohnet says.An agent does something on your behalf. But principals andagents incentives are not always completely aligned, and theresasymmetric information.

Traditionally, academics have linked trust to risk tolerance,since it involves taking a risk. Instead, Were saying that risk-taking when the agent of uncertainty is nature is very dierentfrom when the agent is another person, Bohnet asserts. Afarmer, for example, faces natural risks like weather and soil

conditions. But there are also social risksspeculativebubbles, HIV infection, terrorismwhere other

people produce the uncertainty.Bohnet and Zeckhauser have been run-

ning two games, now with about a thou-sand subjects around the world, playingin groups of 30 at a time. They are two-person games, a variant of the classicPrisoners Dilemma. In the rst game,Player A can choose a safe alternativeor choose to trust Player B, who can in

turn choose an option that rewards bothof them more than the safe alternative, or

a second option that brings even greater win-nings for Bbut less than the safe optionwould have given A. In other words, thegood (i.e., trustworthy) B player will takethe win-win alternative, while the selsh Bwill maximize his own outcome at As ex-pense. When the researchers ask subjectsplaying A, What percentage of good peoplewould there have to be in the room [of 15 po-tential B players] before you would be willingto trust this stranger [the B player]? the an-swer has consistently been 50 to 80 percent.

The second game has the same rules as therst, except that an urn containing 100 blue

and green marbles takes the role of Player B. The urn is a proxyfor an impersonal force, such as nature. If a blue ball is randomlychosen, B selects the trustworthy win-win alternative; if agreen ball, the selsh one. The researchers then asked A players,What percentage of blue balls would the urn have to contain foryou to be willing to take this risk? A rational money-maximizingpersonone who cares only about outcomeswould give thesame answer to this question as to the analogous one in the rstgame. But when playing with nature, respondents generallypeg the gure at 30 to 40 percent, far lower than in the rst game.People are less willing to take risks when confronted with an-other person than when confronted by nature, Bohnet explains.Trust is not only about willingness to take risks, but about thewillingness to be betrayed.

By comparing the dierence between Minimal AcceptableProbabilities in the rst and second games, the researchershave been able to distinguish risk aversion from betrayal aver-sion. The nature game establishes a baseline level of risk aver-sion, but the game with a human Player B introduces the addi-tional possibility of betrayal. Thus, the gap betweenpercentages on the two games gives a rough index of betrayalaversion. In the United States, Switzerland, and Brazil, the be-trayal aversion dierential is 10 to 20 percent. Zeckhauser andBohnet have also played the games in the Persian Gulf region,with subjects in Kuwait, Oman, and the United Arab Emirates.(They are the rst social scientists to run economic experi-

Games of Trust and Betrayal

GE

TT

Y I

MA

GE

S


Harvard Magazine 95

tion of stimulus and response. Neo-classical economics posits a directrelationship between the object andthe choice made. But in behavioraleconomics, the choice depends onhow the decision-maker describes the ob-jects to himself. Any psychologistknows this, but it is revolutionarywhen imported into economics.

We are vulnerable to how choices are described, Wanner ex-plains. Advertising is a business that tries to shape how peoplethink about their choices. Neoclassical economics can explainads only as providing information. But if the seller can invest inadvertising that frames the choice, that frame will skew thebuyers decision. The older economic theories depend on the idea

that the successful seller will produce a better product, the mar-ket will price the product correctly, and the buyer will buy it at aprice that maximizes everyones interestthe market is simplywhere the buyer and seller come together. But once you intro-duce framing, you can argue that the buyer may no longer be act-ing entirely in his own self-interest if the seller has invented a

frame for the buyer, skewing the choice in favor of the seller.Then, the model of the market is not simply buyersand sellers coming together for mutually benecial

exchange, Wanner continues. Instead, the ex-change between buyers and sellers has aspects of azero-sum game. The seller can do even better if hesells you something you dont need, or gets you tobuy more than you need, and pay a higher price forit. The classical welfare theorem of VilfredoPareto was that markets will make everyone as

well o as they can be, that the market distribu-tion will be an ecient distribution that maximizes

welfare. But once you introduce framing, all bets areo, Wanner says. A zero-sum game between buyer and

seller clearly does not maximize everyones welfare, andhence suggests a dierent model of the marketplace.

There are many political implications. We have had 30 years ofderegulation in the United States, freeing up markets to worktheir magic. Is that generally welfare-enhancing, or not? Wan-ner asks. Framing can call that into question. Everyone agreesthat theres informational asymmetryso we have laws that en-sure drugs are tested, and truth-in-advertising laws. Still, thereare subtle things about framing choices that are deceptive,though not inaccurate. We have the power of markets, but theyare places where naive participants lose money. How do we man-age markets so that the framing problem can be acknowledgedand controlled? Its an essential question in a time of rising in-equality, when the well-educated are doing better and the poorlyeducated doing worse.

Its a question that behavioral economics raises, and, withluck, may also be able to address. The eclipse of hyper-rationalEconomic Man opens the way for a richer and more realisticmodel of the human being in the marketplace, where the brain,with all its ancient instincts and vulnerabilities, can be bothpredator and prey. Our irrationalities, our emotional hot-but-tons, are likely to persist, but knowing what they are may allowus to account for them and even, like Odysseus, outwit tempta-tion. The models of behavioral economics could help design a so-ciety with more compassion for creatures whose strengths andweaknesses evolved in much simpler conditions. After all, Theworld we live in, Laibson says, is an institutional response toour biology.

Craig A. Lambert 69, Ph.D. 78, is deputy editor of this magazine.

ments in the Gulf region, and will go toSaudi Arabia in March.) In these coun-tries, betrayal aversion is markedlyhigher, with a dierential in the 30 to 40percent range. Many in this area saythey are willing to trust only if 100 per-cent of the people are trustworthy,Bohnet reports.

She had an enlightening experiencewhen teaching negotiation and decisionanalysis to a group of government ministersfrom the Persian Gulf region in a Kennedy Schoolexecutive-education program. I started the class byasking them to recall a time when they lost trust in someone,Bohnet recalls. One minister said, Trust is not an issue for us. Wenever trust. What a beginning! It opened up a very interesting dis-cussion. A minister said, We cannot dare to trust because we maylose face. I would never come to a meeting and put something onthe table that other people could decline. The meeting-before-the-meeting is absolutely critical in the Gulf, because being let down isterribly humiliating.

Trust has other policy implications. Social capital, per capitaincome, economic growth, and political stability all have positivecorrelations with trust in a society. Trust is a generally goodthing, says Bohnet. And nations deal with breakdowns of trust indierent ways. In the Western world, especially the UnitedStates, contract law builds on the notion of damages or ecientbreach, Bohnet says, meaning that someone who breaches acontract must compensate their counterpart. But if people are re-ally betrayal averse, damages wont satisfy them, because whatthey are concerned with is the fact of betrayal. U.S. contract lawfocuses on decreasing the material cost of betrayal, but what be-trayal aversion asks for is to decrease the likelihood of betrayal,which causes emotional hurt. In Islamic law, which seems to en-courage building trust by personal relationships rather than legalmeans, damages play a much smaller role than in the West. In ad-dition to differences in law, there obviously are other contributingfactors. For example, group-based social organization, typical inthe Gulf but not in Western countries, is based on long-standingrelationships. This substantially reduces the likelihood of betrayaland thus, the social uncertainty involved in trust.

Educational messages may be doomed. They do not resonate. Voters tune out or forget things that are inconsistent with their beliefs.

A n d r e i S h l e i f e r


Documents

Marketplace of Perceptions