Extensions of the Analysis of Choice "[stock market] is high because of the combined effect of indifferent thinking by millions of people, very few of whom feel the need to perform careful research on the long-term investment value of the aggregate stock market, and who are motivated substantially by their own emotions, random attentions, and perceptions of conventional wisdom."Lottery: A tax on people who are bad at math. (a bumper sticker)

Imperfect Information and Uncertainty

Behavioral Economics & Peculiarities

Introduction

Economists, until fairly recently, accepted without question the assumption that people are rational, but then some economists and psychologists started watching people, and conducting experiments, and they noticed some not-so-rational choices. Consider the Ultimatum Game in which two people are given a sum of money, say $100 - with one catch. One person is given the task of dividing the $100 between the two and the other decides whether to take the offer. If the second person takes the offer, then they distribute the money as specified, but if it is rejected, then no one gets any money. The rational person would accept any distribution since they get some money, but often offers are rejected. Or sign up for the experiment designed to see if sexual arousal affects behavior. No real surprise here - people do not tend to make the same choice when they are sexually aroused, so that calculator may not function the same at all times.

Needless to say, many economists cling to the rational model, but even former Federal Reserve Chair Alan Greenspan, a true conservative economist, found himself questioning the rationality assumption. Greenspan identified "irrational exuberance" as the factor driving the stock market boom in the 1990s, and by 2002 the Nobel Prize in Economics was awarded to Daniel Kahneman and Vernon Smith, two of the leaders in the behavioral economics "revolution."₁The difference between the two views is nicely described below.

While economists usually assume that agents are well informed, that their preferences are well ordered and stable, and that their behavior is controlled, selfish, and calculating, psychological research indicates that people's judgments are biased and their preferences malleable and unstable. Further, people can be impulsive, short-sighted, trusting, and vengeful; they often have mistaken intuitions about their behavior; and they frequently effect outcomes they themselves view as bad.... we [should] think of decision makers not as faulty economic agents but as fundamentally different creatures from those envisioned by classical analysis.₂

It turns out that while individual behavior may not be rational, it is often predictable, and in this unit we examine the behavioral economics research to gain some insight into the peculiarities of our behavior. We will also look at the implications this has for the design of public policy and introduce a relatively new concept - libertarian paternalism - that could bridge the ideological divide separating the liberals and conservatives that so often leads to gridlock.

At the center of behavioral economics is the idea that there are two decision-making systems that run parallel to each other. In the first system (System 1), which originates in the oldest part of the brain, choices are based on perception and intuition, they are made quickly, effortlessly, automatically and without much thought, and there is little learning taking place. You do it because you have always done it that way, you do not think much about it, and you are unlikely to change or learn.

In the second system (System 2), the one rational choice economists must have had in mind, choices are slower, more controlled, rule-based and reasoned, and there is more chance for learning. You do it because you have thought long and hard about the choice, what your goals are, and how this choice will affect your attainment of those goals, and as new information comes in, you will have the flexibility to change.

And in between we have the world in which many of our decisions are made, a world where Kahneman believes, "people rely on a limited number of heuristic principles which reduce the complex tasks of assessing probabilities and predicting values to simpler judgmental operations. In general these heuristics are quite useful, but sometimes they lead to severe and systematic errors.".

To see how this dual-process decision process works, answer quickly the following questions. 

1. A bat and a ball cost $1.10 in total. The bat costs $1 more than the ball. How much does the ball cost? 

2. 5 machines working 5 minutes produce 5 widgets, how long for 100 machines to produce 100 widgets?

It turns out that many initially answered ten cents and 100 minutes, both of which are wrong. The problem is the initial response came from System 1 and was not "checked" by System 2, because a simple check would have given answers of five cents and 5 minutes. Or, think about the problem you will have if you see an octagonal red sign with the words GO printed inside, or a door that needs to be pushed, but has a handle on it. The automatic part of the brain will send one message - stop and pull, while the reasoning part of the brain will eventually say - Go and push - and this has not been lost on marketers.₃

The mobile-payment technology can create a desensitizing and seductive purchase experience, said James Katz, director of the Center for Mobile Communications Studies at Rutgers University. “The more people think about a purchase decision, the more likely uncertainty creeps in,” he said. “One frame of mind is you’re helping create in consumers’ mind a source of pleasure, and enabling them to fulfill that pleasure.₄

But this desire for safety wasn't a rational calculation. It was a feeling. Over the past decade, a number of major automakers in America have relied on the services of a French-born cultural anthropologist, G. Clotaire Rapaille, whose specialty is getting beyond the rational--what he calls "cortex"--impressions of consumers and tapping into their deeper, "reptilian" responses.” ₅

We begin the unit by looking at situations where imperfect information can affect decisions and markets’ performance. This is followed by a section in which we examine some of the heuristics, those not-so-rational regularities in our behavior, and then look at the implications of this behavior for public policy. As you work your way through the unit, keep in mind that the acceptance of non rational behavior seriously weakens the case for the unfettered power of markets and limited government intervention - and opens the door for firms to exploit this behavior - the topic for the next unit.

Economics of Imperfect Information

In our discussion of markets the only information ever mentioned was price, and all participants readily knew that information. Unfortunately, this is often not the case in the world of imperfect information where people ask waitresses for meal suggestions, buy a product after watching a commercial staring some celebrity in which we hear little about the specs on the product, and wear a suit to go on a job interview. It is also a world where buyers and sellers are potential rivals prone to sending out erroneous information; where a buyer would not want to reveal how much he wanted a car and where a seller really does not want to let you know everything about the car they are trying to sell. If this sounds like the world in which you live and make choices, read on. 

Limited information

In the real world in which we all live, information is costly to assemble and often, rather than do the research necessary to make informed choices, we look for short cuts. One of those short cuts is the use of signals. Rather than doing the hard research you look for some easily recognizable "sign" to guide you through the choice. For example, if you were in BIG trouble, would you hire a lawyer in a fancy suit and spacious office, or one with a plaid jacket and a second floor walk-up? If you were a recruiter for a company, would you offer a job to a student with an average cum (2.5) when there is a candidate with a 3.8 cum? Would you be influenced in your car decision by an ad offering a 100,000-mile warranty?

The answer to all of these questions is often yes, although in each case the decision is based on limited information. We really want a good lawyer to win your case, not someone with nice taste and a fancy desk. We want to hire someone who will add value to your team, not someone who managed to get a high grade point average in courses. We want a car that has a proven track record of good performance, not one with a warranty. So what were we doing? We were looking at signals - and as you can see below, we are not alone.

Does Christian Dior “overcharge” when it sells a handbag for $13,000? That depends on how you look at it.  If you see the handbag is a few pieces of stitch leather, the price is grossly inflated.  If you see it as a source of heady self-worth – a passport to exclusive club – than it’s hard to say what price would be too high.₆

Not all signals are, however, created equally, and if signals are to be credible, then they must be costly-to-fake. For example, if you see Rolex offering a lifetime warranty, or a car company offering 100,000-mile warranties, you know it costs BIG bucks to advertise, and if their products actually were not reliable, they would not be able to stay in business. These ads serve as good signals to consumers because they would be costly to fake. Korean auto companies were counting on this in 1998 when they began their 100,000-mile warranties on their autos. The Korean cars had been in the US for a while, but there were persistent rumors about the poor quality of the cars - the same thing we heard when the Japanese imports started to arrive three decades earlier. In fact Korean’s strategy was based on Chrysler's earlier experiment with extended warranties when it was on the verge of bankruptcy, and you can expect to see the Chinese auto imports following similar strategies as they try to break into the US market. Big ad budgets, promises for refunds for defects, and service guarantees are all means by which companies signal that their claims are credible, because a potential buyer realizes how much has been invested by the seller in the product and how big the losses would be if the product was not good. It is much the same thing as the lawyer signaling success with the wardrobe, and the student is signaling value with the 3.8 cum. Keep this in mind as you head out there into the job market. Those career counselors’ "dress-for-success" advice is simply an example of you sending a nonverbal message to prospective employers.₇

Some of those messages also illustrate the full-disclosure principle that states if some individual stands to gain by advertising a positive quality of the good or service, then others will be “forced” to reveal the quality of their product even if it is not as good. For example, assume that MG, a major auto manufacturer, advertises its product as having the best consumer rating. Dorf, meanwhile, would like to ignore this since it comes in second in the rankings. Would it volunteer that information? Probably because if it did not then potential buyers would assume it was hiding something and would value it along with all of the other car companies. To avoid this it would advertise its ratings even if they were not first – and then the third ranked company would do it…

And keep this in mind when you head off to that job interview. The US government has made efforts to eliminate discrimination in the labor market, so it has restricted questions an interviewer can ask. For example, in an interview you cannot ask someone about their marital status and plans for having kids. Most firms would not want to hire someone with the prospect of losing them when they decided to stay home and raise kids, and marital status could be used as a signal for that possibility. If an interviewer used marital status as a signal, then the single woman would be in the best position so you would expect a single woman to offer this information even if there was no question. So the government may have outlawed the questions, but you are likely to find that job candidates will offer that information.

The importance of signals is also reflected in the concept of conspicuous consumption, a term introduced by Thorsten Veblin in The Theory of the Leisure Class. According to Veblin, who was writing during the gilded age of the late 19th century, the wealthy spent lavishly on expensive and wasteful goods and services to "prove" their success. How many of those big SUVs are driven because the family needs their functionality, and how many of those Porsche owners ever get to experience the thrill of the speed? Not many I suspect. Conspicuous consumption, such as your SUV or Porsche, is a way of signaling the world that you have arrived, which is why Virginia Postel, in The Substance of Style, suggests conspicuous consumption is not a sign of success, but rather a sign of belonging to a relatively poor group - a wannabe. Domestically, she contends that conspicuous consumption explains why African Americans spend 25% more on jewelry, personal care and apparel, and less on education, entertainment, and home furnishing. Since people tend to lump people together by race, conspicuous consumption is more important for those from lower income groups - African Americans. Internationally, this is why demand for luxury items is soaring in poorer countries, with China set to account for a full quarter of world demand by 2014.₈ It is also why the US, a symbol of success in the world, could have both a direct and indirect affect on climate change. If we are rich and we want SUVs, then we will damage the environment with our exhaust and that of others we can expect to emulate us to "prove" that they too have arrived. This is especially important if "in a post ideological world, who you are is defined less by what you think - if you think at all - than by what you purchase and eat."₉

Imperfect information and its impact on decision making also sheds light on herd behavior that is central to an understanding of speculative bubbles and fads. The tendency to imitate others is not a new phenomenon, nor a product of modern advertising. Machiavelli noticed the tendency when he wrote in 1514: "Men nearly always follow the tracks made by others and proceed in their affairs by imitation," and in the next half millennium not much changed. Philosopher Eric Hoffer, writing in 1955, noted that when "people are free to do as they please, they usually imitate each other. . .. A society which gives unlimited freedom to the individual, more often than not attains a disconcerting sameness."₁₀

I doubt this surprises many of you, and it certainly did not surprise me. It also helps explain why obesity, smoking, and teen pregnancy tend to be contagious; why the decisions of federal judges on three-judge panels are affected by colleagues' votes, with Democrats voting more conservative when the other two are Republicans, and Republicans voting more liberal when the odds are against them; and why college roommates, who are assigned randomly, have a notable impact on student performance.₁₁ This effect has also not been lost on marketers who repeatedly include in their marketing, phrases like "most people prefer" and "growing numbers of people," or on Juliet Schor, author of Born to Buy who wrote:

Teen media depicts a manipulated and gratuitous sexuality, based on unrealistic body images, constraining gender stereotypes, and, all too frequently, the degradation of women.  The dominant teen culture is also rife with materialism and preaches that if you're not rich, you’re a loser. Adolescents are subjected to unremitting pressure to conform to the market's definition of cool

In a world of imperfect information, many individuals "learn" by observing the actions of others - what is referred to as observational learning or social learning. This is why people ask a waitress for information on meal choices, even though they know nothing about the waitress' tastes.  I saw this first-hand on July 4th on a hill overlooking San Francisco Bay. A crowd formed quickly to watch the fireworks over the Bay, but my conversations with those there revealed no one actually knew if you could see the fireworks. Everyone assumed the person there before him or her was right, although soon we all realized we had made a BIG mistake. San Francisco had fireworks, but you couldn't see them from the hill. 

At its simplest, the model is based on the assumption an individual's choice is based on information collected and on the choices of others. How much we rely on our own information and how much on the choices of others depends on the cost of attaining information and the 'stature' of the initial adopter. People are likely to follow the 'lead' of others, which is why MIT, in an effort to reduce drinking, began to run ads stating that most kids did not drink excessively. It also explains why San Marcos, CA was able to cut energy consumption by sending to customers their usage and the average usage figures. Those using more than the average, reduced their consumption, and the reduction was even larger if a smiley face was attached to the bill.₁₂

And very often the "others" are celebrities. This is the "buzz" in entertainment, this is why successful restaurant and bar owners are very good at getting the right people into their restaurants and bars, why initial reviews play such a key role in the eventual success of movies, and why a down-and-out Miami Beach paid to get the "beautiful" people on the beach by subsidizing photo shoots to create the image of luxury and excitement. Miller Lite's high profile athlete endorsement program and Nike's signing of Tiger Woods to gain entry into the golf market, both worked marvelously and helped the companies attain rapid increases in market share. What we are not certain of is why it works - either because we assume they know better, or because we want to be like them. This concept also explains why two management gurus would secretly purchase 50,000 copies of their book in 1995 from the stores monitored by the New York Times bestseller list and why the revitalization of Times Square languished until Disney announced its plans to invest in the area. And then there is Dr. Usmani’s “swarm model” that he is selling to supermarkets. RDFs are placed on shopping carts that signal how many others have purchased the products you are passing in the aisles, which can increase sales without having to give people discounts that would lower profits. ₁₃

Entertainment offers additional examples of the power of suggestion. In an online experiment, 14,000 participants were divided into two main groups and given lists of songs they could download and rate. One group was given a list of songs with no information, while the other group was given information on how often a song had been downloaded. In theory, the rankings should have been the same in the two groups, but they were not. The most popular were much more popular among the group that had download information. More importantly, the group with the extra information was subdivided into 8 groups that got download information on only their group. What was determined was that the hit songs in the subgroups were different, suggesting that it could have been a small difference in the beginning that was magnified.₁₄ The "moral" of the story, get that buzz out early.

Asymmetric information

In addition to people having limited information, there is also a problem with asymmetric information - when one party in a transaction has more and better information – because the party with the information advantage can be expected to exploit that advantage. And what an advantage it is, so pursue an information advantage where possible because it is the best protection against losing in the exchange. Asymmetric information creates market failures that can be traced to two types of opportunistic behavior we examine in this section - adverse selection and moral hazard.

Adverse selection occurs when one party in a trade is taken advantage of because they do not know about an unobserved characteristic of the item being exchanged. Sellers of cars know more about the car's problems, the buyers of health insurance know more than the insurers about their health, and firms offering extended maternity leave know less about an individual's plans for children than the job applicants, and in each case markets fail to produce the correct, competitive, efficient outcome. What we find is that when sellers have more information on product quality than buyers, we end up with markets of lemons.

In the case of the autos, buyers expect sellers to inflate the claims about their cars, so buyers will respond by assuming people with good cars do not sell them, and they will not pay the full value for a good used car. Sellers of good used cars, meanwhile, decide not to sell their good used cars because they cannot get a decent price, and we end up with a used car market with a bunch of "lemons." Because of asymmetric information, a market for good used cars fails to function effectively despite the fact buyers and sellers looking to exchange good used cars exist and would benefit from such a market. This also helps explain why new cars lose their value so quickly. When you drive the new car home from the showroom you have certainly not created $3,000 or $4,000 of wear-and-tear on the car, but this is how much the market price falls since potential buyers assume the worst when a low-mileage used car hits the market.

Adverse selection is also a considerable problem in the insurance industry. In the case of health insurance, the market will likely attract a disproportionate number of people with health problems since those with medical problems would see insurance as a good deal and would buy it. This would prompt insurers to charge higher prices for insurance policies, which would discourage healthier people from buying the health insurance, which would lead to further price increases. The result is that a market-based health care system will tend to provide too little insurance at too high a price, which is why in the 2008 presidential election a key piece of the Democrats’ government health care plan was universal enrollment – ‘forcing’ everyone to get health insurance.

The same would be true with auto, home, and life insurance. Safe drivers would opt out of collision insurance, while people careful enough to install smoke detectors might choose not to buy home insurance. As for life insurance, the "insurers have set premiums high because they know people who expect a long life will snap annuities up as a hedge against outliving their savings, while people who expect to die soon won't."₁₅

In the case of a firm offering generous maternity benefits, those benefits would likely attract people with the intentions to get pregnant. This would raise the costs of doing business for those firms offering the benefits, eventually driving them out of business because and this desirable benefit would cease to exist. Another example is a government sponsored Hope Scholarship proposed during the Clinton administration that would give students scholarships to those in college who would then promise to pay 1.5% of their future earnings to the government as repayment - a rate set by the government to generate enough revenue to finance the program based on the published data for the average income of college educated workers. It sounds plausible, but those who expected to make above the average would not enroll, paying the tuition instead, so adverse selection would leave the average income of participants too low to support the program.

Asymmetric information, in this case where the seller has the better information, also makes it possible for firms to employ price discrimination - a topic we will examine in more detail in the next unit. American automakers, who were in big trouble in 2009, were masters at this. They marketed essentially the same car as different models - a Chrysler and Plymouth, a Ford or Mercury, or a Chevrolet or Pontiac - and by doing so they were able to charge different prices. The same is true with appliances. When buying appliances there appears to be quite an array of appliance brands to choose from at the local store, but if you look under the sheet metal you will find that most were produced by a few manufactures. There are clearly extra costs associated with stamping different names on the machines, but they are more than compensated for the additional revenues generated by the ability to charge different buyers different prices. What remains to be seen is if this strategy proves less successful in the future, as more information is available on line.

Moral hazard refers to situations where the informed participant in a transaction takes advantage of the uninformed one by not revealing certain actions, or where behavioral changes occur after a contract is signed that damage the other party. Buying insurance will probably alter a person's behavior, prompting the insured to exhibit riskier behavior because he/she is covered by insurance. Once you get car insurance you might be more likely to drive a little faster, and once you sign up for health insurance you might be more likely to play a potentially dangerous sport or show up at the hospital for a cold.

A similar situation would be found in the financial sector. When is the last time you examined the books of the bank you have your checking account? I doubt you have, which makes sense because you know you have deposit insurance that guarantees your money. As a result you pay less attention to the financials of the bank - you adopt riskier behavior. This has become a BIG issue in international economics in recent years as some question the policies of the IMF that guarantee loans of poor countries. The fear is that this will encourage them to borrow irresponsibly which would increase the likelihood of additional financial problems. This was precisely the issue at the heart of the financial bailout in 2008, when you heard many warning about the consequences of bailing out home buyers or lenders who had acted irresponsibly because it would send a "message" to others that irresponsible behavior was OK since they would be bailed out too.

Now let's look at potential ways to deal with the problems of adverse selection and moral hazard. There are two general approaches to the problem. First, you try to reduce opportunistic behavior such as the efforts of the sellers of autos to inflate the claims they make about their cars, or the health problems a job applicant suppresses to get the job with generous health benefits. Second, you try to eliminate the asymmetry in information. Bad results come from bad information, so maybe we can "force" the participant with the information to reveal it to the less-informed participant. Here are a few strategies that have been adopted to reduce the impact of adverse selection and moral hazard.

Reducing opportunistic behavior

1. Restrict choices. Rhode Island, as well as many other states, does this when they require every driver to have auto insurance.₁₆Without the universal coverage law you would expect to see too few good drivers with insurance and too many bad ones with insurance. This policy should lower the cost of insurance by eliminating adverse selection. This was an important distinction between the Obama and Clinton proposals for national health insurance in the 2008 presidential campaign. Clinton wanted mandatory enrollment, similar to what was enacted in Massachusetts and what is common in Europe because this would "force" those with low costs into the program so average costs would be lower.  
2. Require waiting periods for insurance. Have you ever lost your phone and then called to sign up for insurance? Maybe you have not done it, but others have, which is why insurance companies often require a waiting period to discourage this type of applicant.
3. Require a co-pay or high deductible for insurance. To discourage applicants with health problems from applying for a job, an employer could impose a high co-pay and raise the share of the medical expenses paid for by the employee. Auto insurance companies could raise the deductible on policies so drivers would pay a greater share of the cost of repairs. This would address the adverse selection problem by discouraging bad drivers from applying for insurance and it would address the moral hazard problem by encouraging drivers to drive more safely.
4. Require rigorous physical effort. To discourage those applicants with health problems from applying for a job, an employer could specify that all jobs involved some extensive physical work so those with physical limitations would not apply.
5. Restrict certain behaviors: Let's look at a professional baseball player who is interested in skiing. We know there are real risks associated with skiing, but if you had insurance against any injuries, you might be inclined to take up skiing as a hobby because there would be no risk. In this case the insurer would attract those with dangerous habits, so to avoid this adverse selection, the insurer could deny coverage for any injuries from skiing or skydiving.

Eliminating the asymmetry in information

1. Screening: The under informed party in a transaction can use screening to gather information. The most obvious example would be the requirement of prior medical or driving records on all insurance applications since this would certainly reduce the information asymmetry. This is a big problem in health insurance since insurance companies will choose not to supply health insurance to those with pre-existing problems leaving them uninsured. Insurers would also want to know family history and any hobbies that might pose risks to the insured. Buyers, meanwhile, could also use screening when making choices. How many of you have bought something because of a review you read, or because of a seller's reputation?

2. Signaling: The informed party in a transaction can use signals to provide more information to the under informed party. For example, you will signal your skills by showing up on time to a job interview, or getting a doctor to sign off on your good health before you buy health insurance, and firms signal the quality of their product with warrantees. A good example of the latter is the certified pre-owned car program. In recent years there has seen a shift from buying to leasing cars that has left dealers with large inventories of relatively new, pre-owned vehicles they must resell, but potential buyers will underestimate their value. The dealers, needing a way to gain credibility for their claims of reliability, began the certified pre-owned car program that provides the buyer of a used car with a warranty, a signal to the potential customer that this is not a lemon. This is also why on EBay you look for seller ratings, the closest you can come to a warranty on what you are buying.

3. Product liability laws: A seller might think carefully about selling an unsafe product if they would be held liable for any safety problems associated with their product. A good example of this are the "lemon laws" that require companies to inform customers if the car has ever been classified as a lemon. By doing this the government is reducing the asymmetry in the buyers' and sellers' information.₁₇

4. Standards and certification: The government, business groups, or consumer groups provide a metric for evaluating products. When insulating a home you have the R-value, and when borrowing money for a mortgage on your house you get the effective annual rate that allows comparisons across mortgage types, and when investing in financial assets they have a rating.₁₈

We have now seen how rational decision-makers may choose to operate with less than perfect information that creates an additional problem known as the winner's curse.

Winners curse

Imagine the situation facing oil companies considering their bids on offshore oil leases. The US government, representing us, would set up the auction so the companies could bid for the rights to drill for our oil and then sell it. This would be a common value auction where the auctioned item should be of equal value to all bidders - the value of future profits in this case. The problem is that the companies do not have perfect information on the costs, revenues, and profits from the fields, and the auction winner would be the company that most overestimates the value of the lease. Similar situations would arise when studios bid for scripts, athletic teams bid for players, and networks bid on Olympics. Another example would be the spectrum auctions, such as the one in 1994, where the FCC auctioned off 99 licenses for wireless service. As with oil, the carriers would not know in advance what the profits would be, so the winner would be the highest bidder and would have to make a profit at a price above the average estimate of future profits. An interesting strategy appeared in one auction when Pacific Telesis, who wanted the California market, took out ads in local papers where potential rival bidders were located and hired economists to lecture to these companies on the problem of winner's curse. It worked - bids were low and Pacific Telesis got a good deal on the California market.

In recent years we have seen auctions for pay-per-click online advertising, a simple concept mastered by Google. When you type in a search term, Google provides a list of sites for you to search for information on the left side of the page plus a list of advertisements on the right side of the page. If you click on one of the ad links that takes you to a company webpage, then Google gets paid a certain price per click, and the company gets a potential customer. The auction comes in because many companies want to be placed on the ad list when you type in your search term, so they bid on each search term and the highest ranking on search engine goes to the highest bidders.

And finally, I found myself intrigued by the idea proposed by some that there was a winner's curse in marriage. It actually is not that much of a stretch if you think of everyone interested in marriage competing in a market for a partner. The person who gets married will do so because they have most overestimated the value of the partner, and so they are likely to be disappointed, just as the oil company that wins the oil lease auction.

Uncertainty, Risk, and Expected Utility

So - let’s say you decide to spend the time and assemble all of the important information. You checked the weather data to determine the day with the greatest chance of sunshine and scheduled your party on that day. There is still one problem – it might rain. It turns out that often in our lives we have to make choices without certainty. We go to a movie and are not certain we will like it; pick a restaurant and are not certain we will get a good meal, and we select a stock without knowing what will happen to its price. In each case we have alternative outcomes to which we can associate some probability: the chance of a good meal is 80%, or the chance of a stocks' price rising is 55%. Now we are going to look at how we analyze choice when we have uncertainty.

Decision making in an uncertain world is similar to gambling, and at the center of both is the concept of expected value. To see how it works, let's look at Max's fund-raising walk. Max knows the results of the walk depend very much upon the weather. If it is a good weather day, Max's walk will bring in $10,000, but if it is a rainy day then the walk will not cover the costs and Max's walk would lose $2,000. Max also knows the probability of a nice day in October based on historical weather data is 70%. With this information we can calculate the expected value of the walk - how much Max could expect, on average, to earn from the walk. The expected value is simply the weighted sum of the outcomes, with the weights being the probability of the outcomes.

Expected value = .7*$10,000 + .3*(-$2,000) = $7,000 - $600 = $6,400

There is always a chance it will rain and Max will lose, but if he were to run the event many times in October there would be more sunny days than rainy days and the average take from the event would be $6,400.

As a second example, calculate the expected value of a lottery ticket where you have a 1% chance of winning $200 and a .5% chance of winning $1,000. The expected value of the lottery ticket is $7. If you bought hundreds of lottery tickets you would expect on average to win $7 on each ticket, so it is very unlikely you will ever see anyone offering you this deal.

Expected value = .01*$200 + .005*($1,000) = $2 +$5 = $7

These calculations are pretty straightforward, and they lead to additional questions: Should you hold the walk, and how much would you pay to play the lottery? The answers depend upon one's attitude toward risk, and to see how it works we need to look at how someone treats a fair bet - one where the expected value = 0. Would you be willing to pay $7 for the lottery ticket with the expected value of $7? There are three possibilities?

Risk averse - You would rather have the $7 with certainty than the expectation of the $7 so you would not buy the lottery ticket for $7. You would not buy the ticket unless you paid less than $7 to play. Risk averse people will always refuse a fair game. They would never agree to a coin-flip game where they win $100 with a tail and lose $100 with a heads.

Risk seeker - You would rather have the lottery ticket with the expected value of $7 than the $7 with certainty, and you would be willing to buy the lottery ticket for less than $7. Risk seekers people will always accept a fair game. They would agree to a coin-flip game where they win $100 with a tail and lose $100 with a heads.

Risk neutral - You are indifferent to the $7 with certainty or the lottery ticket with an expected value of $7. You would pay just $7 for the lottery ticket.

What are you? Surveys indicate people do not like risk, and fortunately there are ways to reduce risk.₁₉ First, you could just avoid choices involving risks, but that will not always work. Second, you could do your homework and get yourself good information, which is what investors thought they were doing in the Bush II boom. Unfortunately, the rating agencies did a very bad job of rating risk and rated even very low-grade assets highly. Third, you could diversify, which is at the center of so many investment strategies. For example, if you were concerned with investing and rates of return, you could choose to own one stock or you could choose to hold a variety of stocks. There would be less risk with the portfolio of stocks because the odds are pretty good that not all stocks would move in the same direction. The odds are good, at least if all of the stocks do not move in the same direction (a negative correlation), that the losses of a declining stock would be offset by gains in others. This is precisely the concept behind the development of mutual funds that allows small investors the opportunity to diversify their portfolios.

And let’s not forget insurance. You could insure yourself against risk, and to see how insurance works let’s look at a simple example of health insurance, the topic of a heated debate in 2009. In our simple world there are two groups of people – those who will have a healthy year with annual health care expenses of $1000 and those facing an unhealthy year whose average health care expenses are $10,000. In this situation, if 80 percent of the population is expected to be healthy, then the expected health care costs for a year are $2,800.

Expected value = .8*$1,000 + .2*$10,000 = $800 +$2,000 = $2,800

Faced with these numbers, how much would a firm charge for insurance? If the insurance company offered a fair game, it would charge $2,800, but the company would never offer such a deal. First, the company would have to build in some profits to its price so it would need to charge more than $2,800. Second, even though the probability figures are true averages, the expected health care expenses for you and your grandfather would be quite different. Based on the track record of millions of people, the probability of your grandparent having a bad health year is much higher, so you would expect to see adverse selection in the health insurance market. Young, healthy people would decide not to buy insurance, while older, less healthy older people would buy it so the expected cost would rise. If you assume that we end up with only 60% of the health insurance buyers being from the healthy group, then the expected costs would rise to $4,600 so the insurance cost would need to be higher than this figure.

Expected value = .6*$1,000 + .4*$10,000 = $600 +$4,000 = $4,600

And we are not done yet because we may also expect to see costs rise as some of those in the healthy category take risks because they have insurance – an example of moral hazard. Let’s assume that this moves 10% from the healthy to unhealthy category. In this case the expected value of costs would be

Expected value = .5*$1,000 + .5*$10,000 = $500 +$5,000 = $5,500

What started as $2,800 in expected health care cost has now turned into a health insurance company needing to charge more than $5,500 to cover expenses and earn a profit.

There are two important implications of this math. First, it makes financial sense for many young, healthy people to self-insure themselves for small losses since they know the expected health costs are far less than the insurance premium. The same would be true with auto insurance. It probably makes sense to get an insurance policy with a high deductible so you would be paying for any smaller losses out of your own pockets. Second, an important feature of the 2009 health care reform proposal was mandatory coverage because this would eliminate the adverse selection effect and thus lower the expected health care costs. In essence the plan is doing “forcing” young, healthy individuals to buy the insurance, which means the pool of insured will have more healthy people in it so the same costs can be spread over more people.

Now before you decide to swear off marriage because of the winner's curse, and change your health insurance coverage, let's look at some additional peculiarities observed by behavioral economists, much of which is based on experiments in which the researchers observe the actual choices people make in a variety of circumstances - and many of those observed are students. More recently, the emerging field of neuroeconomics where researchers observe MRIs of the brain while people are making choices has complemented this research. 

Behavioral Economics and Peculiarities in Individual Choices

In the previous section we examined choices people made with imperfect information, and generally people are comfortable with those choices. You may know the fancy suit may not be what you really care about in your choice of a lawyer, but you are likely to be comfortable with the choice. Now we examine some choices people make that are not based on imperfect information, but rather on imperfections in the choice mechanism. Richard Frank, at a Federal Reserve Bank conference, described two sets of choices we make - those without regret and those with regret. In the previous section you did not have regret for your choices, but here we will talk about some behaviors you will likely regret. Furthermore, I guarantee you will find examples of behavior you would call your own, and if you accept the evidence presented here, you end up with a very different view of behavior and a different view of public policy. In Frank's view, government intervention should be focused only on those choices where individuals feel regret. So let's get with it - and we'll start with a few I find most relevant for students.

Self-control and resisting temptation

We start here because I believe it helps explain why there are so many students who tell me they want an A and so few who actually do the needed work; why so many people claim to want comfortable retirements and so few actually save enough; and why obesity is so prevalent in our society despite the fact that very few would ever claim they want to be obese. The reality is, "The road to hell is paved with good intentions." There is often a large discrepancy between our words and our actions, which raises questions about the assumption of rationality if people are not doing what they want to do. What we have here is a problem of self-control, or more precisely, the lack of self-control. Neuroscientists, with the help of brain scans, have found the self-control problem arises because the brain is in conflict with itself: one part saying go for it and one saying do not - and too often the conflict is won by the weaker side. This conflict would not come as a surprise to those familiar with the story of Adam and Eve whose inability to refrain from eating that apple brought havoc on their lives.

The good news is that there are things we can do to alter the outcomes of the conflict. In fact, this is at the heart of the self-help literature that focuses on implementing a plan rather than formulating the plan. An early example can be found in Homer's Odyssey when Ulysses, recognizing the limitations of his will power, had his crew tie him to the mast to avoid the temptation of the Sirens.

While we may not tie ourselves to the mast too often, you do need an effective commitment device – something that helps you keep that promise. For example, consider the situation faced by someone who will need to write the IRS a check of $12,000 next year. There are two options: deposit money each week in a savings account that would earn interest, or increase the federal tax withholding so more would be automatically taken from the weekly paychecks? A rational person would never choose to walk away from free money, but this is precisely what they do when they withheld more taxes. This makes sense only for someone compensating for the lack of self control who knows those weekly deposits will not all be made. This lack of self-control is what drove many of my parents' friends to enroll in Christmas clubs with "forced" deposits each week, and in recent years it has driven some compulsive gamblers to put themselves on a "self-exclusion" list baring them from casinos, and if they are caught, they agree to allow themselves to be arrested and have their winning confiscated.₂₀

If you want to quit smoking, you may want to stay away from situations where there will be smoking, and if you do not want to sleep through the alarm clock, put it on the other side of the room so you will need to actually get out of bed to get it. For those who want to lose weight, try the following. Get a friend who also wants to lose weight and agree to meet whenever either of you requests a weigh-in, and if one of you turns out to have exceeded the pre-assigned weight, that that person pays $500.

And what about those grades? If you want to get those A's on your transcript, there are many things you can do help make that wish a reality including establishing a commitment device. In one experiment, college students enrolled in a class that required three semester papers were divided into three classes. In class #1 the students could choose any dates and then stick to them, in class #2 they were told there were no deadlines and all papers could be passed in on the last day, and in class #3 they were given due dates spread across the semester. Not surprisingly, if you believe in the powers of procrastination and the limits of self-control, at the end of the semester the grades were best for those in class #3, and worst for those in class #2. The middling performance of class #1 showed that the students who pre-committed to deadlines did do better than those with no deadlines, and that within class #1 it was the students who chose no deadlines that pulled down the average. So do not bet on your self-control to get that paper in at the last moment. Set some deadlines - your commitment device - and stick to them. For those procrastinators, and I know you are out there, you might consider some variation of a strategy used by one professor who was trying to get his student to finish a dissertation on time. The student wrote the professor a number of $100 checks at the outset of the semester and each week the dissertation was late the professor cashed another check.

Opportunity costs and money illusion

People have a tendency to ignore certain costs in their choices, which should be no surprise based on your calculations of the cost of a year of college, which would not happen in a world of rational choice. Neither would you find the following behavior observed by economist Alan Krueger. In the 2001 Super Bowl, the one in Miami where the Ravens crushed the Giants, hundreds of Super Bowl tickets were allocated by lottery to the thousands of people seeking tickets. Krueger tracked down the winners and asked them if they hadn't won, would they pay $3,000, the average price of scalped tickets. 94% said no. Then Krueger asked the winners if they would sell the tickets for $3,000, and 92% said no. Just as students overwhelmingly fail to acknowledge opportunity cost when calculating the cost of college, the winners of the tickets failed to account for opportunity cost in their choices. By choosing to use the ticket go to the Super Bowl, they were giving up the opportunity to sell the ticket for $3,000 - so going to the game is really costing them $3,000. They just did not realize it since it was an opportunity cost and not an out-of-pocket cost.

For those who do not like sports, consider the situation faced by a friend who likes wine, but not enough to ever spend more than $30 for a bottle. Ten years ago, however, she bought a case of wine for $5 a bottle, and as luck would have it, today it is worth $200 a bottle. The wine merchant told her that the vintage has reached its peak value and offered her $100 a bottle to buy it, so I was initially surprised when she turned down the offer and broke open the wine. Here was a friend who said she would never pay more than $30 for a bottle of wine, and she was paying $100 for a bottle. She paid far more for the wine than she said she would spend because she didn't get opportunity cost right.

One last question: Do you think it is fair to charge a friend or family member interest on money they borrow from you? Many find it is unfair to charge interest because it costs nothing to lend out money – or at least that is the story. In fact if you were lending out $10,000, you could not invest it and earn $s. If interest rates were 5%, then you would not earn the $500, so it would actually be costing you $500 to lend the money.

A second problem people consistently have with financial transactions is they suffer from money illusion- they do not do a good job of adjusting prices to account for inflation. To see the problem, consider the following choice. Option 1 is you buy a house for $100,000 and sell it 5 years later for $127,628, while option 2 is you buy a house for $100,000 and sell it 5 years later for $115,927. Which would you prefer? I suspect you would do what most do and pick option 1 because the sales price is about $12,000 higher, but you could be very wrong. The problem is inflation was not taken into account in the choice - and it should have been.

To see how inflation matters, let's assume in option 1 that inflation averaged 7%, while in option 2 prices rose by 1% a year. If you do the math, the true buying power of the two alternatives look quite different. The value after the adjustment for inflation is actually higher in the second situation. In the first situation the house's value rises 2 percent slower than prices, so you would be behind by 2 percent. For example, suppose you wanted to buy a $100,000 car today with the money from the house, but you were holding off till you sold your house in five years. Unfortunately, because the price of the car increased with inflation at 7% a year, the car would cost $140,255 so with the proceeds from the home sale ($127,628) you could not afford the car ($140,255). Clearly you would be worse off in five years. The situation is very different in the second option where you would be ahead by 2%. Returning to the house and the car that both cost $100,000 today, in five years you could sell the house for $115,927 while the car would cost you  $105,101. In this situation you could buy the car and pocket some money - probably to pay for the gas.

Sunk Costs

"We have already spent $80 on the $100 project, so we must finish it." You have certainly heard people express this type of opinion, and many of you probably find nothing wrong with a behavior economists find very troubling because it is not rational. One of the "rules" of the rational choice model is that costs you have already incurred, sunk costs, should not affect current decisions, just as the choices of profit-maximizing firms should not be affected by fixed costs. The theory is quite simple - rational choices should be based on a comparison of benefits and costs associated with the choice, and since sunk costs are unaffected by the current choice, they should be ignored. In the choice of completing the project, the $80 should be irrelevant; what should matter is the benefit associated with the last $20. If completion of the project creates only $10 of value, then it makes no sense to do it even if you have already spent $80, but often people’s choice is affected by that $80 sunk cost. It turns out, based on considerable research that sunk costs affect many decisions. Here are a few examples of the experiments used to determine the extent to which sunk costs matter.

1. In a study of theatre goers at a Midwestern university, described in Why Smart People Make Big Money Mistakes, season ticket holders were randomly selected to receive one of three different prices for their tickets. It was observed that the highest rate of attendance occurred in the group that paid the highest price, which suggests that people behaved as if sunk costs mattered.

2. Assume a person belongs to tennis club and needs to pay an hourly fee of $15 for an indoor court in addition to the annual tennis membership of $150. Play on the outdoor courts is free. Consider the following two scenarios and tell me if you would expect similar results.

Scenario 1: A person has scheduled and paid for an indoor court on Saturday at 2:00 PM, and when the time arrives it turns out to be a beautiful day, one on which playing outdoors would be preferred. Would this person choose to play on the outdoor courts?

Scenario 2: It is time to play tennis on a beautiful Saturday at 2:00 PM. Would a person choose to play on the outdoor courts?

It turns out that the answer depends very much upon whether the $15 had been paid, but it should not have mattered. The rational choice model predicts the decision to play on the outdoor court is the "rational" decision regardless of whether or not the fee had been paid, but the results show that sunk costs do affect behavior.

3. Imagine the plight of Monica who has just bought a $200 pair of shoes and finds that they hurt her feet, even after they have been broken in. What will she do: keep wearing them or get rid of them? Now that you have your answer, consider Monique’s plight: she has the same shoes with the same problems, but she was given the shoes as a gift. What will she do?

It turns out that people are much more likely to ditch the shoes when they were a gift, which makes no sense in the rational choice model because the source of financing the shoes has no bearing on the pain from the shoes.

So, make sure you do not make these mistakes in your choices - and now let's turn to look at how uncertainty affects choices and prospect theory that really got behavioral economics started.

Prospect Theory

At the center of the behavioral, or psychological approach to decision making, is prospect theory, first introduced in 1979. As early as the 1950s there had been questions raised about the validity of the rational choice model.₂₁ Herbert Simon suggested the bounded rationality model that emphasized the limits people have in formulating and solving complex problems, but it really was prospect theory, with its experimental emphasis, that raised the profile of the rational choice theory's critics. It was initially seen as a rebuttal to the expected-utility model of choice that could not explain the fact that people simultaneously gamble, suggesting they are a risk taker, and buy insurance, suggesting they are risk averse? There must be something else behind people's choices than the expected-utility, and after considerable research on people's behavior, Daniel Kahneman and Amos Tversky (K&T) proposed prospect theory as an alternative explanation of people's behavior.₂₂

To see the difference between the two models, consider the following situation. Mary comes home and finds a rebate check of $100 on her water bill and a speeding ticket for $80. The expected utility model suggests it is a good day since her wealth has increased. K&T, however, think people treat these two events differently – they separately assess the affect of the two events and in the assessment they weigh the loss (-$80) more than the gain (+$100). Rather than making a decision based on the value of a net increase in wealth of $20, they believe people make decisions based on the values they separately attribute to the $80 loss (ticket) and the $100 gain (rebate). Furthermore, they believe the weight of evidence indicates people tend to value losses twice as much as gains.

The essential features of prospect theory are captured in the graph below. If you look to the right of the origin, you are looking at how gains are valued, while to the left of the origin you have the valuation of losses. There is no symmetry here: the curve is steeper for losses suggesting losses are more painful than the gains are rewarding.

Asymmetric Value Function

To see how this might work in practice, let’s look at the change in health care plans at a local company. The existing plan paid 100% of all medical costs at a price of $200 a year. The proposed plan costs $100 and has an $80 deductible – each person is responsible for the first $80 of medical expenses and then the company covers all other expenses. If you look at it as a simple math problem, the new plan is better since everyone can be expected to save a minimum of $20. So how could people object to the change? If they looked at the $100 savings in premiums as a gain and the additional medical expenses of $80 as a loss, then the new plan would not be the preferred plan.

Another feature of prospect theory can also be seen in the asymmetric value function diagram above - the importance of framing. Because the function is concave in gains and convex in losses, the value of a gain of $40 and a gain of $60 would be greater than a gain of $100. So if you are selling products or giving gifts, segregate the gains and the recipients will be happier. As for losses, you should combine them since the value of a loss of $80 is less than the value of the loss of -$20 and -$60.

In fact there are many additional implications of the model, a few of which are listed below.

1. Investing: Why is there a BIG equity premium for stocks? People can invest in stocks or bonds, but because stocks' values vary more they are riskier so investors would expect a higher return on stocks to invest in them. This is called an equity premium - the additional return on stocks needed to get us to take on the additional risk. The problem is it is much bigger than you would expect based on the data, but this could be explained by the fact people are averse to losses.  Because investors holding stocks experience more losses than investors holding bonds, those investing in stocks will need to be compensated with higher returns. Shiller, in "Human Behavior and the Efficiency of the Financial System," describes an experiment in which potential investors were given either a table of annual rates of return data for stocks or a simple table that had the 30-year average rate of return. The investors that saw the list with the annual ups and downs invested a greater percentage of their pensions in bonds.

2. Investing: Why do investors tend to hold falling stocks too long and rising stocks too short? Prospect theory suggests investors will hold on to falling stocks because they HATE to lose, while they will cut their gains short when stocks are rising because they do not value the gains as highly. In one study reported in Camerer, it was observed that investors held losing stocks about 25% longer than they held gaining stocks.

3. Elasticity of demand: Why is there an asymmetry in responses to price increases and price decreases? If you look at price increase as a loss for the consumer and price cuts as a gain, then as a result of people's loss aversion, they will be more responsive to price increases, hence demand is more elastic when prices rise.

4. Consumption & Saving: Why are people's consumption expenditures less responsive to bad news than good news? According to O'Reilly, a $100 increase in the stock market adds $3.5 to spending, while a $100 decline pushes spending down by $5 - $6. They do this for two reasons. First, because they are loss averse. Second, because they are risk takers with the loss, they would rather take no change today and a chance of a big change next year to a moderate change this year.

It has also been observed that people have trouble with probabilities. People tend to overweight outcomes that are considered certain relative to outcomes that are merely probable, so certainty increases the aversion for losses and increases the value of gains. People also consistently misestimate probabilities - overestimating very small probabilities and underestimating larger probabilities. Peter Bernstein, in Against the Gods: The Remarkable Story of Risk, reports that people's estimates of the probability of deaths were way off, with people overestimating the likelihood of a death from unnatural causes and underestimating the likelihood of deaths from natural causes. And, when probabilities are large, people go with the larger probability, while when the probabilities are small, people go with the larger reward. A good example of how this behavior has affected marketers can be seen in those state lotteries. State lotteries can increase the size of the payouts at the same time they have reduced the odds of winning because the traditional "rational" player would see the expected value as unchanged and would be unaffected, but those who tend to ignore small probabilities, or overestimate them, would see the bigger payouts as a better game and play more.

Status quo and endowment effects

Another effect that links current choices with previous choices is the status quo effect, or the tendency people have to fall in love with what they have and hate to get rid of it. In "Why Johnny Can't Invest" there is a good example of the effect. Say, for instance, that your Uncle Dudley has died and left you $10,000 in government bonds. He has also bequeathed your brother (who has the same appetite for risk that you do) $10,000 of stock in a risky Internet company. You can't both be content with such drastically different investments. So what will you do? The odds are high that a year later neither of you will have done nothing. Or think about your TV viewing. How many times have you watched a show because it just happened to be on the same channel after a show you had watched? It probably is not as strong an effect now that we have remotes, but it must still exist given the talk you here about the importance of the lead show on prime time TV.

Another great opportunity to see the power of the effect arose in happened when New Jersey and Pennsylvania established slightly different insurance programs. In New Jersey the default insurance policy - the one you get unless you change it - had a low premium and no right to sue. If you wanted the right to sue you would need to buy extra insurance. In Pennsylvania the default insurance policy had a high premium and the right to sue. If you wanted to give up the right to sue, you could give up that right for a lower premium. In effect you have two policies that provide the same coverage, the only difference being the default. It turned out, however, that between 75 and 80 percent accepted the default - and unless you can make the case that PA and NJ residents have dramatically different perspectives on insurance and risk, then you are seeing a bias toward the status quo. Schwartz describes another experiment, one involving the purchase of a new car. In this case some people were offered a new car with the choice of taking an additional $500 in options while the others were offered the car including the $500 options in the price and the option of not taking the options. While you would expect the same share of both groups to take the options, what was found fewer people ended up with options if they needed to add them to the purchase.

The status quo effect is in many respects very similar to what has been called the endowment effect. Rational choice theory suggests that the value attached to something should not be affected by whether we happen to be a buyer or seller, but experiments have repeatedly shown that value depends upon the side of the transaction on which you find yourself. In one common experiment, mugs are allocated randomly to some people in a group. Those who have them are asked to name a price to sell their mug; those without one are asked to name a price at which they will buy. Usually, the average sales price is considerably higher than the average offer price.23 Once people take "ownership" of something, they tend to resist giving it up because they overvalue the status quo or maybe because they see giving it up as a loss. This may very well be why oriental rug companies allow people to take home rugs to "test" because once they are home the people will become "attached" to them and be less likely to give them up. It is also why you see so many 30-day money back guarantees.

Another interesting experiment indicating the existence of the effect took place at Duke University. Duke University has a very small basketball stadium and the number of available tickets is much smaller than the number of people who want them, so the university has developed a complicated selection process for these tickets that is now a tradition. Roughly one week before a game, fans begin pitching tents in the grass in front of the stadium. At random intervals a university official sounds an air-horn, which requires that the fans check in with the basketball authority. Anyone who doesn't check in within five minutes is cut from the waiting list. At certain more important games, even those who remain on the list until the bitter end aren't guaranteed a ticket, only an entry in a raffle in which they may or may not receive a ticket. After a final four game, Carmon and Ariely called all the students on the list that had been in the raffle. Posing as ticket scalpers, they probed those who had not won a ticket for the highest amount they would pay to buy one and received an average answer of $170. When they probed the students who had won a ticket for the lowest amount they would sell, they received an average of about $2,400. This showed that students who had won the tickets placed a value on the same tickets roughly fourteen times as high as those who had not won the tickets.24

Taken together the sunk cost, endowment, and status quo effect make people pretty resistant to change.

Framing and anchoring effects

One of the assumptions of rationality is that decision makers are not affected by irrelevant information, but repeated experiments have shown this is often not the case because there are framing effects in decision-making. Choices are affected by how a situation is presented (framed), so "[p]references can be malleable, context dependent, and inconsistent even though decision makers are thoughtful, serious, and engaged."25 We can see this by examining the problem specified in Kahneman's Nobel lecture that appears below. In the experiment some people were given the first choices, while others were given the second.

Choice 1: Imagine that the United States is preparing for the outbreak of an unusual Asian disease that is expected to kill 600 people. Two alternative programs to combat the disease have been proposed. Assume that the exact scientific estimates of the consequences of the programs are as follows:

• If Program A is adopted, 200 people will be saved.
• If Program B is adopted, there will be a one-third probability that 600 people will be saved and a two-thirds probability that no people will be saved.
• Which of the two programs do you favor?

Choice 2: Imagine that the United States is preparing for the outbreak of an unusual Asian disease, which is expected to kill 600 people. Two alternative programs to combat the disease have been proposed. Assume that the exact scientific estimates of the consequences of the programs are as follows:

• If Program A is adopted, 400 people will die.
• If Program B is adopted, there will be a one-third probability that nobody will die and a two-thirds probability that al people will die.
• Which of the two programs do you favor?

If you do the math, these are equivalent choices and thus the choices people make should be the same - but they are not. What Kahneman and others find is that when given the first choice, a substantial majority favor Program A, while Program B is favored by the majority when given the second choice. Kahneman explains this by noting, "outcomes that are certain are over-weighted relative to outcomes of high or intermediate probability. ... Thus, the certainty of saving people is disproportionately attractive, and the certainty of deaths is disproportionately aversive." 

There is also the issue of anchoring. It turns out how you ask something matters a lot, which is what we saw with the framing issue. For example, what percentage of the world's population lives in North America? Researchers have conducted experiments where people are asked numerical questions such as this, but first they were given another task unrelated to the answer. Some were asked to identify their favorite number while others were asked to spin a roulette wheel that generates a random number. What they found out is their favorite number or the number on the roulette wheel influenced the answer to the population question. As the number on the wheel increased, so did the estimates of answer to the question. Another indication of the power of anchoring can be seen in pricing decisions, and you should keep this in mind the next time you find yourself haggling over price. When people are uncertain about a price, what researchers find is that the final price and initial offer are positively related - the final price rises when the initial offer is higher.₂₆

Proof that anchoring does in fact exist, or at least that many retailers believe it works, can be found in the prevalence of "sale" signs. How often did you find yourself saying that a sales item was too good to pass up because the price had been dramatically reduced? You were influenced by the original price that was also listed on the slip and acted as an anchor to you.

While you are thinking about your shopping behavior, think about how often you have done something like the following. When considering the price of an item you compare it with another similar item that influences your decision. If you are looking at a watch that sells for $200, you are more likely to buy it if the other watches in the case are $500 than if they are $50. The other watches are the anchor against which you make your decision, which is why Schwartz reports that a high-end catalog seller's sales of a $279 bread making machine almost doubled when a $429 deluxe brand appeared in the catalogue. All of a sudden the original bread maker looked like a bargain because the deluxe brand became the anchor.

Anchoring also provides some useful tips for those in the fundraising or restaurant businesses. If you are running a fundraising campaign for which you print cards with suggested gifts, make sure to put in big numbers because this will tend to act as an anchor and raise the average level of giving. In a restaurant, meanwhile, make sure you have on the menu at least one very high-priced item because this will tend to raise the tab. And if you are Starbucks entering a market where there are already established players such as Dunkin Donuts, then make sure your branding does not allow coffee drinkers to compare Starbucks prices with those in Dunkin Donuts.

Those interesting in real estate might appreciate the next two examples.₂₇ A group of realtors were taken for a tour of a home and given a sheet of comparables - the sales prices and specific attributes of a number of houses recently sold in the neighborhood. The only difference in the packets given to the realtors was the estimated sales price on the house they toured. After touring the house and reviewing the document they were asked to provide an estimate of the market price. What they found was that the estimated price was higher for those realtors with the packets in which the house's price was higher suggesting the listed price, which was arbitrary, acted as an anchor in estimating the price.

Another example cited by Shiller was the interpretation of price-earnings ratios. He notes that in the 1980s US investors believed the Japanese market was overpriced, while similar price-earnings ratios in the US in the 1990s were viewed with less suspicion. The explanation was that in the 1980s US investors used the lower US ratios as an anchor when assessing the Japanese stock market.

Psychophysics of perception - or truth about relativity

Another peculiarity of behavior is that people's choices are influenced by context. People do not know what they want until they see it in context. For example, let's look at a shopping question: Will you drive across town to save $5 on a $50 radio - and would you drive across town to save $5 on a $500? The rational choice model would say that there should be no difference - if you were willing to travel for a $5 saving, you should do it both for the $50 radio and the $500 TV. But experiments show a difference with the $5 mattering less as the price rises. This has been described as an example of the psychophysics of perception - people's ability to identify differences in a signal is directly related to the intensity of the signal. For example, you might be able to distinguish a 10 from 20 watt light bulb, but not a 100 and a 110 watt bulb, and in this case you could distinguish the $5 fee on a $50 purchase, but not on a $500 one.

This seems to be a favorite characteristic for marketers looking to enhance revenues.₂₈ In the first, a seller of a high and low priced item would increase revenues by adding a third option in the middle. For example, consider the subscription options offered by The Economist magazine. The Print-only option is clearly only a decoy and you would expect no one to pick it since the Print-and-Internet only option costs the same - which is precisely what the students did. More importantly, in Game B where you take out the decoy you would expect similar results to what you observed in Game B, but they were not similar with the high priced option chosen far less than in Game A. The decoy had worked.

Game B	Game A
Internet-only $59	Internet-only $59
	Print-only $125
Print-and-Internet only - $125	Print-and-Internet only - $125

Representativeness

The "problem" of representitiveness is very close to what we would call stereotyping, and to see it in operation, consider your answer to the question: "Is a 6'10" man more likely to be a basketball player or a salesperson?" Many people are tempted to pick the professional basketball player since this is where you see the highest concentration of tall people. The problem is there are very few professional basketball players and a lot of sales people, so it is very possible the odds of the person being a basketball player are very low. For example, if there are 400 professional basketball players in the country and 20 million sales people, and 20 percent of the basketball people are taller than 6'10," as are .01 percent of salespeople, then the number of 6'10" + basketball players is 80 and the number of 6'10" + salespeople is 2,000. Based on these data it is far more likely that the 6'10" person will be a salesperson despite the fact that most assume that he is a basketball player. 

Another way of looking at representativeness is to think of it as the result of a tendency on the part of people to underestimate the power of randomness. The classic example of this is a study of basketball fans that tend to believe in the existence of streak shooting. Most fans, including me until I came across the study, believe a player will hit his next shot if he hit his last one, or better yet, his last few shots. So get the ball to the shooter with the hot hand is perceived as a good strategy. The problem is there is no evidence to support it in the data. The evidence clearly indicates the last shot has no impact on the current shot, but if you listen to the announcer of a basketball game you would never believe it was not true.₂₉

Regression to the mean

While many believe in the Sports Illustrated jinx, there is a good alternative explanation - regression to the mean. To understand the problem, just think about how one gets on the cover. Athletes only get on the cover because they are having a remarkable year, one that is well above average. Once you accept this you realize the statistical odds are that the "star" will not be as remarkable next year so that the cover person can be expected to have a less successful year.

 Do you believe in the Sports Illustrated jinx that assures anyone appearing on the cover of the magazine to a sub-par year next season?

Let's look at another example, one you may be familiar with. What is likely to happen tomorrow after someone is praised, or scolded, for their performance? The law of averages suggests performance will return to the mean, so after the person is "scolded," you should expect the behavior will improve tomorrow - again just because of the law of averages. The problem is that people tend to forget about the law of averages and attribute the behavioral change to their policies. Unfortunately, there is an asymmetry her because "scolding" of bad performers can be expected to be followed by a better performance, while rewarding exceptional performance will be followed by a return to more normal, worse behavior. The policy of praise does not appear to breed continued superior performance.

Public Policy and Libertarian Paternalism

Given these "imperfections" in human behavior, what should be the role of policy makers? There are two dominant views - the paternalistic liberal left who believe government should play an active role in managing the economy and the libertarian conservative right who believe the government should get out of the way of businesses and people who are doing what they want to do. Is there any way to bridge this divide? At the Federal Reserve Bank conference the researchers and policy makers generally supported the idea of "libertarian paternalism" that offers some hope for bridging the divide between the left and right and breaking the gridlock we see on so many policy issues.

A starting point is that because people's choices are influenced by legal and organizational rules, it is impossible for government not to affect choices. Government by definition is paternalistic, and once you accept this, then the issue is one of designing policies that allow freedom of choice while at the same time prompting people to make the best choices.

In Nudge we get a good example of this situation. We know that obesity is a problem in the US, and we know many young students get school lunches, so can we / should we do something about the obesity problem by altering the choice architecture of the school lunch program? In this particular situation the question is: how should we display the food because it does matter, and the school has control over how the food is displayed. Here are some options.

1. Choose the food order at random
2. Try to arrange food to get kids to pick same foods they have chosen in the past which would probably keep them happy now
3. Display prominently the foods on which the suppliers have given the best prices, which would lower costs of the lunch program.
4. Arrange food to make students best off, so you put the healthier foods in the front of the display.

The potential appeal of libertarian paternalism is that you must give people choices, but you could stack the deck and steer people in ways that improve their lives by carefully selecting the defaults. In this case the 4th option would do that.  As for when this approach would be most effective, there are some guidelines suggested by the authors of Nudge. If we know people sometimes exhibit behavior that may not be in their best interest, then maybe we can use public policies to "nudge" them toward better behavior. This is most likely to be important when people are making choices that are made difficult by the fact that the benefits are in the future and the costs no, when the choices are made infrequently (choice of school or mate), and when there is limited feedback. We saw earlier the impact of the choice architecture in people's insurance choices in NJ and PA, which is why the "choice architecture" is so important.

An important example of the impact of the choice architecture on people's behavior is the Save More Tomorrow Plan designed to increase people's savings. Consider the following options for a savings plan at work. In the first, which is what you find in most jobs, when a person gets a job they make an appointment at which they specify how much they would like taken out for savings, and that rate stays in place until the person makes the effort to change it. In the second, your employment comes with you already enrolled, and the rate of savings will be automatically raised in the future. If you did not want to do either, you could contact your employer and change the rates. The bottom line is both plans would allow you to choose whatever savings you wanted, where they differ is in the default, and this matters. The second plan will generate higher savings rate, as can be seen in the results of New Zealand's "KiwiSaver" plan designed to increase retirement savings.

Two other policies that reflect the power of the status quo bias are organ donation, and pension annuities. With regard to organ donation, if the default on a policy was for donation of organs, you can be assured the share of people donating organs would be substantially higher, even though there is no reason to believe the groups had different views on donation.

As for pensions, individuals have a choice of earmarking the money for themselves or setting up a joint-survivor-annuity that guarantees that their spouse will get the pension if they die. In the US the law was changed in 1974 so that the default would be the joint-spouse-guarantee was the default, and the percentage choosing this option increased by about 25 percent - even though there was no change in preferences. There is also the issue of the allocation of your assets within a pension fund. When you are young the advice is to take risks and invest in stocks, while those who are nearing retirement are advised to be more conservative and invest in bonds. You could have a program where the default was the allocation scheme would needed to be changed, or the default could be that the allocation would be set to change at specified dates in the future. Given the status quo effect you can be assured that the results would be quite different. And in the future we can expect to see more debate on the privatization of SS that will give people increasing control over their retirement. The outcome of any privatization will depend greatly upon the choice architecture, so you should check out Sweden's experience with privatization.30

So we have now observed one of the pieces of that rational model that was believed to bring us close to that utopian state, and we have found that assumption of rational consumers to be a bit off the mark. The significance of this should not be underestimated since it was consumer sovereignty that was at the center of this utopian view - we would have an economic system that would give people exactly what they wanted at the least cost. The loss of the rationality assumption weakens the case for an unfettered market system and opens the door for enlightened public policy - and one of the directions to think about for future policies would be what has been called libertarian paternalism. We will continue on with this theme now as we shift to those other decision makers - firms - and see what happens when we relax some of the other simplifying assumptions we have made.

---

1. For an overview of their work see Daniel Altman's article, "A Nobel that Bridges Economics and Psychology," New York Times October 10, 2002. 2.

 2. Eldar Shafir at the Federal Reserve Bank of Boston's annual conference "How humans behave: Implications for economics and economic policy." ,

3. Another example is the idea that discrimination between "in" group and "out" group members is "wired" into people's unconsciousness, and this affects their decisions greatly, even though consciously they would disavow any thought of discrimination. ["Unconsciousness raising"] This is what would appear to be behind the much discussed "Bradley Effect" in the 2008 presidential campaign where white voters might have trouble voting for Obama despite the fact they said race did not matter.

4. Matt Richtel, "At Starbucks, Songs of Instant Gratification," NYT, October 1, 2007

5. Gladwell, “Big & Bad,” The New Yorker, January 12, 20044. 

6. Elisabeth Eaves, “The Lap of Luxury,” New York Times October 25, 2005 5.

7. The good news is dress-for-success pressure will tend to decline with age, since by then there will be an established track record and potential buyers will not need to rely as heavily on signals. The same would be true for a lawyer who has been practicing law in town for a number of years. The prospective clients in town will know about the lawyer's success rate and this is likely to be more important than the clothes worn. For a new lawyer coming to town, meanwhile, the trappings of success will be very important since this may be all people know about the lawyer and these will be the signals upon which they will base their decisions. This helps explain why people in large metropolitan areas tend to spend more on expensive clothing than people in smaller communities. As the amount of information falls, the importance of signals increases.

The extent of the conspicuous consumption effect causes problems for luxury goods makers who must weigh the costs and benefits of expanding their market to the not-so-rich. Think about Louis Vuitton, which can charges fabulous prices for its bags because of their exclusiveness, contemplating the expansion of the product line to the those lower down the income distribution. They would gain additional, lower income customers, but risk alienating the super wealthy who no longer got any "value" from being seen with the same bags that a secretary would be wearing.

8. Virginia Postrel, The Substance of Style (2007) and "Conspicuous Consumption," The Atlantic Monthly, July / August 2008 and "Conspicuous Consumption in China: Luxury's New Empire," The Economist 2004

9. Roger Cohen, "Premiumize or Perish," NYT September 15, 2008

10. Bikhchandani, Hirschleifer, and Welch (1998)

11. Richard Thaler & Cass Sunstein, Nudge, Yale University Press, 2008

12. ibid.

13. "Swarming the shelves, “The Economist” 11/11/2006)

14. Duncan Watts, "Is Justin Timberlake a product of cumulative advantage?," NYT April 15, 2007

15. As we look forward there is a potentially enormous problem with health insurance created by the advances in genetic testing. "Governments thus face a choice between banning the use of test results and destroying the industry, or allowing their use and creating an underclass of people who are either uninsurable or cannot afford to insure themselves…. Indeed, genetic testing may become the most potent argument for state-financed universal health care." “Testing Times,” The Economist, October 9, 2000

16. To understand the nature of insurance, let's look at a simple probability problem. An important historical example would be the risk associated with outfitting one of those old sailing ships - like the one Magellan sailed around the world and others sailed to the Spice Islands. The potential rewards were enormous, and so were the risks. Assume each ship that sails has an 80% chance of returning and paying the owner $20,000 and a 20% chance of being lost at the cost of $50,000, enough to bankrupt the owner. The expected value (profit) of the trip is $20,000*.8 - $50,000*.2 = $6,000. In this case the owner faces a good chance of hitting a home run, and a small chance of losing everything - a perfect place for an insurance policy. If the owner had two ships that were sent sailing together so the risks were identical, then you have two possible outcomes - they both return or they both get lost - and the expected return calculation appear in the table below. To calculate the potential losses and gains, you use the rules of probability. In the case of these two ships sailing together, there is a 20% chance of losing it all so the expected loss from losing both ships would be .20*-$50,000 = -$10,000 per ship, or -$20,000 for the two. There is also a 80% chance of both ships returning so the expected gain from both ships returning would be twice the expected gain for one ship of  .8*$20,000 = $16,000 - for a total gain of $32,000. In this case the expected gain is $12,000, while you run a 20% chance of losing it all.

Risk and Return

	Probability	Return
both return	80%	32,000
both lost	20%	-20,000
Net profit		12,000

Now let's make only one change - let's send the two ships in opposite directions so the chance of the ships being lost are unrelated. To calculate the potential losses and gains, you use the rules of probability that indicate the probability of two independent events occurring - two losses - is simply the multiplication of the two probabilities. In the case of these two ships, there is a 4% (.2*.2) chance of losing it all so the expected loss from losing both ships ($100,000 loss) would be .04*-$100,000 = -$4,000 - substantially lower than the expected loss from sending one ship. There is also a 64% (.8*.8) chance of both ships returning so the expected gain from both ships returning ($40,000 gain) would be .64*$40,000 = $25,600. There is also a 32% chance that one ship will return and one lost so the expected return would be .32*-$30,000 = -$9,600. In both this and the one boat scenario the expected return is $12,000, but in the second case the chance of losing it all is only 4% instead of 20%. If you were risk averse then you would pay a premium to avoid the initial situation, which is where insurance comes into the picture. Because people tend to be risk averse, an insurance company playing the probabilities could offer the owners a policy where they would pay a premium so as not to be exposed to the risk. Note: What you have just seen here is the value of diversification, which is why this is a key element in any discussion of investing. For those who are risk averse, a preferred investment strategy would be centered around a diversified portfolio of assets.

Risk and Return

	Probability	Expected gain
both return	.8*.8 = 64%	25,600
both lost	.2*.2 = 4%	-4000
on returns	(.8*.2) + (.8*.2)  = 32%	-9,600
Net profit		12,000

It seems like a win-win situation, except here is where adverse selection kicks in. From what we know about adverse selection, the ship owners who are careful would opt out of insurance and the only ones buying insurance would be those who take gambles. This would mean that the expected losses for the insurance company would be larger than expected, which would be BIG trouble for the insurance company.

A similar problem can occur in the labor market where after someone is hired they might have a tendency to slow down, especially if they are paid a salary or an hourly wage. This is a problem of asymmetric information since the worker has more information on the work effort, and as you would expect, firms have come up with a number of strategies to deal with this problem and motivate workers

Monitoring of employees

Incentivized compensation  - piece-rates or commissions. For example, if you owned a retail store and you were concerned about the lack of effort on the part of your sales people, you might move to a commission basis for your workers so their pay was related to their sales. If you did this, however, it may have a downside since your employees would no longer be inclined to help each other out.  

Efficiency wages - pay workers more than prevailing wage so they feel fortunate to have the job and will be motivated to work diligently

Profit sharing - ties employee compensation to company profits.

17. We can also see adverse selection in the labor market. In Europe where the unemployment and safety net benefits are generally much higher than in the US, you will see the least productive leave the work force to be "taken care of" by the system, and this leaves the most productive in the workforce so you might expect Europe to have higher measures of labor productivity ("Taxing the poor to pay the poor") In the capital market we see adverse selection among borrowers. It is difficult for lenders to get all of the information on buyers, but you can be quite certain that lenders assume borrowers tend to be drawn from the ranks of those with the worst credit - and they are probably correct. This is why you will see a bank ask for BIG collateral on any loan so the borrower has much to lose if the loan is not repaid. The supply of blood offers another example of the effect of adverse selection. Today the supply of blood is based completely on volunteers because when it was based on paid donors, the supply of donors willing to sell their blood was not a random sample of the population. Too often the donor was someone with "bad" blood who needed the money, and this tainted the nation's blood supply.

18. The downside of these could be restrictions on competition and a lowering of average quality if the standard has a high and low classification scheme because all of the ones that were not quite good enough for good will be lowered to meet the bad and / or there will be none made that exceed the good standard. The end result is this drives up prices to consumers because fewer sellers are there and the average quality is reduced.

19. The problem for many researchers is that while people indicate in surveys that they are risk averse, they tend to gamble which raises questions about consistency because they should not take fair bets if they are risk averse - and certainly they should not take bets for which the expected value is less than 0. There are some potential explanations. First, gambling is entertainment so people will pay to play. Second, people could have tastes such that they place a high value on a big potential gain, and third, they simply do not know the probabilities.

20. In addition to lack of self control, one possible explanation for this disconnect between goals and actions would be the existence of positional goods, what economist Fred Hirsch called goods you can never get enough of to satisfy yourself. If you have a car, then it will always be compared with a car with more features and better lines. The same will be true for the schools you attend, the clothes you wear, the vacations you take, and the homes you live in. As a result of this "keep-up-with-the-Jones" effect associated with positional goods, people will save less than they want or should.  This has prompted some to suggest that Social Security should be considered forced savings needed to overcome the effect of positional goods. Without it our lives would be filled with too much "stuff" and too little saving for retirement.

21. If you are looking for a good overview, although it is a bit difficult to work through, you might check out Daniel Kahneman's 2002 Nobel Prize Lecture, "Maps of Bounded Rationality: A Perspective on Intuitive Judgment and Choice." You could also read Kahneman & Tversky, Prospect Theory: An Analysis of Decision under Risk. If there is any question about its significance, they should disappear when you realize it has been ranked as the second most cited paper in economics.

22. Choices among risky prospects exhibit several pervasive effects that are inconsistent with the basic tenets of utility theory. In particular, people underweight outcomes that are merely probable in comparison with outcomes that are obtained with certainty. This tendency, called the certainty effect, contributes to risk aversion in choices involving sure gains and to risk seeking in choices involving sure losses. ... An alternative theory of choice is developed, in which value is assigned to gains and losses rather than to final assets and in which probabilities are replaced by decision weights. The value function is normally concave for gains, commonly convex for losses, and is generally steeper for losses than for gains. Decision weights are generally lower than the corresponding probabilities, except in the range of low probabilities. Overweighting of low probabilities may contribute to the attractiveness of both insurance and gambling. The key insight here is that value is assigned to gains and losses, and when we add the other features specified above we end up with a graph that captures the choice situation - the asymmetric value function.

To get started look at the following options.

Option A: You get $750 with perfect certainty

Option B: You get $500 with 50% chance and $1,000 with 50% chance.

The expected mathematical value of the two are the same [EV (Option B) = .5*$500 + .5*$1,000 = $250+$500 = $750], but that is not true for the values. Option A ($750) would be valued at 85, while Option B would be evaluated at 80 (.5*60+.5*100 = 30+50=80). In this case clearly Option A is the preferred option, so you are risk averse here - you chose the certain outcome. Now that you have it, let's check out the following game.

Option A: You lose $750 with perfect certainty

Option B: You lose $500 with 50% chance and lose $1,000 with 50% chance.

The expected mathematical values of the two are the same [EV (Option B) = .5*-$500 + .5*-$1,000 = -$250+-$500 = -$750], but that is not true for the values. The Option A (-$750) would be valued at -170, while Option B would be evaluated at  -165 (.5*-120 +.5*-210 = -60 - 105 = -165). In this case clearly Option B is the preferred option, so you are a risk taker here - you chose the risky option.  So what does this mean? It means this model of behavior can explain why a person gambles and buys insurance, something the expected utility model of rational behavior could not do. 

23. "Rethinking thinking."

24. Wikepedia. A more thorough discussion of the effect can be found in Dan Ariely, Predictably Irrational, Harper 2008

25. Richard Kopcke, Jane Sneddon Little, and Geoffrey Tootell, "How humans behave: Implications for economics and economic policy.

26. George Lakoff in Don't Think of an Elephant: Know Your Values and Frame the Debate--The Essential Guide for Progressives, writes that one of the problems faced by Democrats in the 2004 election was the Republicans were better at framing than the Democrats. As an example he notes the difficulty of being against "tax relief" and the success of Bush at making multilateralism sound ridiculous when he said in his State of the Union address that the US did not need a "permission slip" to protect itself.

27. These examples come from Shiller's book Human Behavior and the Efficiency of the Financial System

28. These examples can be found in Dan Ariely, Predictably Irrational, Harper Collins, 2008

29. This example can be found in Richard Thaler & Cass Sunstein, Nudge, Yale University Press, 2008

30. This example can be found in Richard Thaler & Cass Sunstein, Nudge, Yale University Press, 2008