Week 15 · Chapter

Chapter 15 — Asymmetric Information, Behavioral Economics & Inequality

Principles of Microeconomics Free sample edition

The module's primary reading · ~50–60 minutes · Objective 8 (analyze market failures — externalities, public goods, common resources, and asymmetric information — evaluate responses such as corrective taxes, cap-and-trade, and private bargaining, and survey behavioral economics and the measurement of inequality evenhandedly), with Objective 1 (use the tools of economic modeling — scarcity, opportunity cost, the production possibilities frontier, comparative advantage, and the gains from trade — and distinguish positive statements (what is) from normative statements (what ought to be)) · Read it first this week; the lecture, the slide deck, the tutorial, the quiz, and the Model Workshop all build on it.

In plain language

Some trades go wrong because one side knows more: the seller knows a car's history, and you get one short drive; economists call that gap asymmetric information (one side of a deal knows something the other does not). This week you will see how that gap can empty a market of its best goods, and how warranties and inspections win the trade back. You will also meet behavioral economics (the study of the predictable ways real people choose differently from the careful chooser in our models), including why a default option can change what millions of people do. Then you will measure how evenly income is shared with one curve and one number. Last, you will look back over the whole term and ask when markets work well, when they fail, and how to weigh what to do about it.

Why you already care about this

Picture Adaeze shopping for a used hatchback on a listings site. She finds one at a fair-looking price, meets the seller in a parking lot, and gets a 10-minute test drive. The seller has driven that car for 6 years. The seller knows whether it stalls on cold mornings, whether the transmission slips on hills, and whether the last repair bill was 40 dollars or 1,400. Adaeze knows none of that, and in her place you would not either. She has to decide how much to offer for a car whose quality she cannot see.

That gap is not a small detail. In 1970 the economist George Akerlof showed that a gap like this one can shut a market down. If buyers cannot tell good cars from bad ones, the owners of good cars may refuse every offer, and only the bad cars are left for sale. Every model since Week 3 assumed that both sides see the same good. This week you drop that assumption, then two more: that every chooser weighs every cost with care, and that the model's efficient outcome is the end of the story.

By the end of this chapter you will be able to:

tell adverse selection from moral hazard by asking whether the hidden thing comes before or after the deal;
work through the lemons model step by step and explain why good cars leave the market;
explain how a signal or a screen can bring the lost trade back;
name four behavioral patterns (bounded rationality, loss aversion, present bias, and anchoring) and the power of defaults, and describe each as a predictable pattern;
draw a Lorenz curve from quintile shares, compute a Gini coefficient by trapezoids, and say what it means;
lay out the inequality debate with positive and normative claims labelled, and sum up when markets allocate well and when they fail.

Carry this sentence with you: traits hide before the sale; actions hide after it.

Where we are

Already covered: Week 1 separated positive statements (what is) from normative statements (what ought to be). Weeks 3 and 4 built the running market and its equilibrium at 6 dollars and 30 units a day. Week 6 measured total surplus, 90 dollars a day at that equilibrium, and set efficiency apart from equity. Weeks 11 and 12 showed how market power costs surplus, and Week 13 explained why wages differ. Week 14 opened the study of market failure with externalities and public goods.
This week: asymmetric information (adverse selection, moral hazard, signaling, and screening), behavioral economics (bounded rationality, loss aversion, present bias, anchoring, and defaults), the Lorenz curve and the Gini coefficient, the inequality debate, and a synthesis of the whole term.
How they connect: asymmetric information is one more way a market can fail, alongside Week 14's externalities; the surplus tools of Week 6 measure what the failure costs, and Week 1's positive and normative line organizes every policy question.

This is the last teaching chapter of the course, and you will close it by putting every week's model side by side. Two ideas carry straight over. From Week 6: a trade creates surplus when a buyer values a good more than it costs the seller, and the efficient outcome makes every trade like that. From Week 1: a positive claim can be tested against evidence, and a normative claim rests on values. You will lean on both all week.

1. Name the gap: asymmetric information

You meet information gaps all the time, and so do the people you trade with. A restaurant knows how clean its kitchen is; you see only the dining room. A job applicant knows how hard they work; the employer sees a one-page résumé. A person buying health coverage knows more about their own habits than the insurer does.

Asymmetric information is a situation in which one side of a deal knows something relevant that the other side does not.

The gap comes in two forms, and the way to tell them apart is timing.

Adverse selection is the problem of hidden characteristics before a deal: when buyers cannot see quality, the sellers keenest to trade at an average price hold the worst goods, which can drive high-quality goods out of the market.
Moral hazard is the problem of hidden actions after a deal: once one side is protected, for example by insurance, it may take less care in ways the other side cannot observe.

The used car is adverse selection, because the car's condition is a fact before the sale. Sections 2 through 4 work through it. Section 5 returns to moral hazard.

Misconception check. Students often think asymmetric information means one side is lying. Nobody in the used-car story has to lie; the seller knows things the buyer has no way to check. The cure: asymmetric information is about who can see what, not about honesty.

2. Set up the used-car market

Akerlof's 1970 paper used used cars to make the idea concrete, and this course turns his story into a small classroom model, Dataset R. Its numbers are illustrative, chosen so that the arithmetic stays clean; they are not prices from any real market.

Half the used cars of one model are in good condition, and half are lemons.

A lemon is a used good, such as a car, with hidden defects that its seller knows about and a buyer cannot see before the sale.

Each kind of car has a value to a buyer and a lowest price its owner will take.

Dataset R: what each kind of used car is worth to buyers and owners, in dollars per car (illustrative)
Kind of car	Share of cars	Worth to a buyer	Owner's lowest acceptable price
Good car	half	10,000	9,000
Lemon	half	4,000	3,000

Read the table one row at a time, as you would a price list. A good car is worth 10,000 dollars to a buyer and 9,000 dollars to its owner, so a sale at any price between those two leaves both better off. A lemon is worth 4,000 dollars to a buyer and 3,000 dollars to its owner, so a sale between those two helps both as well. If everyone could see quality, every car would sell. That is the efficient outcome from Week 6: every trade worth making gets made.

Misconception check. Students often read "lemon" as "worthless." In this model a lemon is still worth 4,000 dollars to a buyer and still worth trading. The cure: a lemon is a lower-quality car whose defects are hidden, not a car nobody should want.

3. Follow the market as it unravels

Now take away your ability as a buyer to see quality, which is Adaeze's position in the parking lot. She knows that half the cars are good and half are lemons, but she cannot tell which one she is looking at.

Worked example — the uninformed offer. Follow the market one choice at a time.

The buyer's offer. A buyer who cannot tell the cars apart offers the average of the two values: (10,000 + 4,000) ÷ 2 = 14,000 ÷ 2 = 7,000 dollars.
The good-car owner's choice. The owner of a good car needs at least 9,000. Because 7,000 < 9,000, that owner refuses to sell.
The lemon owner's choice. The owner of a lemon needs only 3,000. Because 7,000 > 3,000, that owner accepts.
What buyers learn. Only lemons are offered at 7,000. A buyer who pays 7,000 gets a car worth 4,000 and overpays by 7,000 − 4,000 = 3,000 dollars.
The new price. Once buyers expect a lemon, they pay no more than 4,000 dollars, a lemon's value to them. Lemons trade at a price between 3,000 and 4,000.

In words: the average offer is too low for the good cars and generous for the lemons, so the good cars leave and the buyers' fears come true. That process is adverse selection.

Now you can measure what it costs, with the surplus idea from Week 6.

The gain lost on each good car: 10,000 − 9,000 = 1,000 dollars, from a trade that both sides would want and that never happens.
The gain kept on each lemon: 4,000 − 3,000 = 1,000 dollars.

So half the possible trades, worth 1,000 dollars each, disappear. In words: the market still works for lemons, but every owner of a good car keeps a car that some buyer values more, and each of those missed sales wastes 1,000 dollars of surplus. That waste is why economists count asymmetric information as a market failure.

One honesty note: this classroom model has only two kinds of car and buyers who all think alike. Real markets have many grades of quality, and the unraveling is usually partial, with good cars selling less often or for less, rather than vanishing.

Misconception check. Students often take the lemons result to mean that all used cars are bad. The model starts with half the cars in good condition. The cure: the result says good cars can be driven out of a market when buyers cannot tell them apart, not that good used cars do not exist.

4. Rescue the trade: signaling and screening

Markets are not helpless against hidden quality. If you are on either side of the deal, you can act to close the gap.

Signaling is an action by the better-informed side of a deal that would cost too much to fake, such as offering a warranty, which reveals what that side knows about quality.
Screening is an action by the less-informed side of a deal to sort the other side by hidden quality or risk, such as paying for an independent inspection before buying.

A warranty works as a signal because it is cheap for the owner of a good car and expensive for the owner of a lemon. An owner who promises to pay for repairs on a car that will need many of them loses money on the promise. So an owner who offers the warranty is telling the buyer something the buyer could not see.

Worked example — the trade comes back. Suppose an inspection, or a warranty, lets the buyer see which car is which.

Good cars. The buyer values a good car at 10,000 and its owner needs 9,000, so any price between 9,000 and 10,000 dollars works for both.
Lemons. The buyer values a lemon at 4,000 and its owner needs 3,000, so any price between 3,000 and 4,000 works for both.
The surplus. Each good-car sale now adds 10,000 − 9,000 = 1,000 dollars of surplus, the same amount that adverse selection wasted.

In words: once quality is visible, both kinds of car sell at their own prices, and the 1,000 dollars per good car that was lost in Section 3 comes back. A seller with a good car gains from offering a warranty, and a buyer gains from paying for an inspection, as long as each costs less than the surplus it rescues.

You can see the same logic far beyond cars, in places you visit every week. A restaurant may post its inspection rating in the window. An online seller builds a record of reviews. A worker earns a certificate that shows a skill an employer cannot see in a short interview.

Misconception check. Students often mix up which side does what. The cure: the informed side signals and the uninformed side screens. The seller who offers a warranty is signaling; the buyer who hires a mechanic is screening.

5. Spot the hidden action: moral hazard

Now move the hidden thing to after the deal, where you cannot watch it. Ravi rents an e-bike for a weekend through an app. The rental includes full damage coverage, so a scrape against a curb will cost him nothing. He still means to be careful, but he takes the bumpy shortcut he would avoid on his own bike. The rental company cannot see how he rides.

That is moral hazard: once one side is protected, its behavior changes in a way the other side cannot observe. An insured driver may park a little more carelessly. A tenant who will never pay for wear may let a leak go unreported. The hidden thing is an action, and it happens after the deal.

The common responses all leave some of the cost with the person whose care is hidden.

A deductible is the part of each loss that an insured person pays before the insurer pays the rest, a common response to moral hazard because it keeps some cost of carelessness with the person who can prevent it.

Worked example — a deductible. Take a fresh, illustrative case. Mai's phone plan covers a cracked screen, and a repair costs 240 dollars.

Full coverage. Mai pays 0 dollars and the insurer pays 240, so a careless moment costs Mai nothing.
A 60-dollar deductible. Mai pays 60 dollars and the insurer pays 240 − 60 = 180 dollars.
The share Mai bears. 60 ÷ 240 = 0.25, a quarter of the repair.

In words: with the deductible, a dropped phone costs Mai 60 dollars, so she has a reason to buy a case, and the insurer can charge less for the plan. Monitoring works too: a rental firm can photograph a bike before and after each ride, and an employer can check the receipts on a company fuel card.

Misconception check. Students often swap the two problems. The cure is the timing test: a hidden trait that exists before the deal, such as a car's condition, is adverse selection; a hidden action that happens after the deal, such as Ravi's shortcut, is moral hazard.

6. Meet the real chooser: bounded rationality

Every model in this course has assumed a chooser who weighs every cost and every benefit with care, the habit you practiced as thinking at the margin in Week 8. You are that chooser on a good day. Real people get close to that chooser on big, repeated decisions. On many others they lean on shortcuts.

Think of the last time you chose a phone plan from a page with twelve options, each with a different data cap, price, and contract length. Very few people compute the cost of every plan for their own usage. Most pick the plan a friend has, or the one in the middle of the page.

Behavioral economics is the study of the predictable ways real people's choices depart from the fully careful, rational chooser assumed in the standard models.
Bounded rationality is the idea that people make choices with limited time, attention, and information, and so rely on rules of thumb rather than weighing every cost and benefit.

Rules of thumb are often sensible. Comparing twelve plans could take an hour, and an hour has an opportunity cost, as Week 1 showed. The trouble comes when a shortcut leads people the same wrong way, again and again. Because those mistakes repeat in patterns, economists can study and predict them.

Pause and reflect — choose one. Pick either prompt; they count the same.

(a) In two sentences, describe a choice you made recently with a rule of thumb, such as a phone plan, a meal, or a route, and say what weighing every option would have cost you in time.
(b) In two sentences, describe a choice you made with care, comparing every option, and explain why that choice was worth the extra effort.

Misconception check. Students often hear "behavioral economics" and conclude that people are stupid. The patterns in this chapter show up in careful, educated people too. The cure: a behavioral bias is a predictable pattern in how people choose, not a lack of intelligence, and that predictability is what makes it useful to study.

7. Feel the sting: loss aversion

Imagine you find a 20-dollar bill on the sidewalk. Now imagine you lose a 20-dollar bill from your pocket. The amounts are the same, but for most people the loss stings more than the find pleases.

Loss aversion is the tendency for a loss to weigh more in a person's choices than a gain of the same size, a central element of Kahneman and Tversky's prospect theory.

Daniel Kahneman and Amos Tversky published prospect theory in 1979. Their idea changed how economists describe choices under risk: people judge outcomes as gains or losses from where they stand, and losses loom larger.

Loss aversion predicts patterns that a careful chooser would not show. It predicts that you will mind a "surcharge for paying by card" more than a "discount for paying cash," even when the two prices come out the same. A surcharge feels like a loss, and a discount feels like a gain. A seller who paid 90 dollars for a concert ticket may refuse to sell it for 70 even when 70 is more than the night out is now worth to them. Week 8 showed that the 90 dollars already spent is sunk and should not drive the decision; loss aversion helps explain why it often does anyway.

Misconception check. Students often think loss aversion means people hate spending money. It is narrower than that. The cure: loss aversion compares a loss with a gain of the same size, and says the loss counts for more.

8. Spot present bias and anchoring

Two more patterns turn up in almost every budget, maybe yours.

Lucía decides each spring that she will start putting 25 dollars a week into savings next month. Next month arrives and the 25 dollars has a use right now, so the plan slides to the month after.

Present bias is the tendency to weigh costs and benefits that arrive now much more heavily than the same costs and benefits later, so that plans to save or study keep being put off.

Present bias fits a familiar story: a person buys a yearly gym membership full of resolve, then puts off the first visit week after week. It is also why "I'll start the reading tomorrow" is such a common sentence. The careful chooser in our models would weigh next month's cost the same way whether it is decided today or next month. Real people often reverse the plan when the moment comes.

Now picture a jacket with a tag that reads "was 80 dollars, now 50 dollars."

Anchoring is the tendency for a first number a person sees, such as an original price on a tag, to pull later estimates and judgments toward it.

Worked example — reading an anchor. Take the jacket as an illustrative case.

The anchor. The first number you see is 80 dollars.
The framed saving. 80 − 50 = 30 dollars, which the tag presents as money you keep.
The question a careful chooser asks. Is the jacket worth at least 50 dollars to me? If you value it at 45, buying it costs you 50 − 45 = 5 dollars of surplus, whatever the tag once said.

In words: the "was" price changes how the deal feels, not what the jacket is worth to you. Your decision rule is the same one you used in Week 6: buy when your willingness to pay is at least the price.

Misconception check. Students often treat a bias as something only other people have. The cure: these patterns are predictable in most people, which is why a store prints the "was" price on the tag in the first place.

9. Set the default: the power of defaults

Whenever you open a form with an option already checked, someone chose a default for you.

A default is the option that takes effect when a person makes no active choice, such as automatic enrollment in a retirement plan that an employee may leave.

Brigitte Madrian and Dennis Shea studied a real case in a paper published in 2001. One large U.S. firm changed its 401(k) retirement plan so that new employees were enrolled automatically unless they chose to opt out. Before the change, employees had to sign up on their own. Participation rose sharply after the switch, although every employee kept exactly the same right to join or leave.

A careful chooser in our models would not care which box starts checked, because leaving it takes only a minute. Real employees did care. Bounded rationality, present bias, and the sense that the default is the recommended choice all push people to stay where they are put.

Keep the two kinds of claim apart here, as you learned to in Week 1.

Positive: in that firm, automatic enrollment raised participation. That claim can be tested against the firm's records.
Normative: whether an employer or a government should set defaults, and which default it should pick, rests on values: how much to trust people's own choices, and who decides what counts as a good outcome.

Misconception check. Students often think a default removes choice. In the study, every employee could still opt out. The cure: a default changes the starting point, not the set of options, and that small change can move behavior a great deal.

10. Draw the Lorenz curve

Now the chapter turns from how people choose to how income ends up shared. Picture yourself lining up every household in a region from the lowest income to the highest, then cutting the line into five equal groups.

A quintile is one of five equal groups formed by ranking households from the lowest income to the highest, so each group holds a fifth of all households.
A cumulative share is the percent of all income received by one group of households together with every poorer group, the running total that a Lorenz curve plots.
A Lorenz curve is a graph of the cumulative share of income against the cumulative share of households, ranked from poorest to richest; the farther it sags below the line of perfect equality, the more unequal the distribution.
The line of perfect equality is the straight diagonal on a Lorenz chart along which every fifth of households receives a fifth of income, the benchmark the Lorenz curve is compared with.

The curve is named for Max Lorenz, who proposed it in 1905. This course uses Dataset S, a set of illustrative quintile shares. They are not any country's figures; a real figure appears in this course only with a dated Census Bureau source.

Dataset S: share of income by household quintile and the running total, in percent (illustrative)
Quintile	Share of income	Cumulative share of households	Cumulative share of income
Poorest fifth	5	20	5
Second fifth	10	40	15
Middle fifth	15	60	30
Fourth fifth	25	80	55
Richest fifth	45	100	100

Worked example — the running total. Add each fifth's share to the total so far.

Poorest fifth: 5 percent.
Through the second fifth: 5 + 10 = 15 percent.
Through the middle fifth: 15 + 15 = 30 percent.
Through the fourth fifth: 30 + 25 = 55 percent.
All households: 55 + 45 = 100 percent.

In words: the poorest 60 percent of households together receive 30 percent of income in this illustrative region, and the richest fifth alone receives 45 percent. Compare the two ends: the richest fifth's 45 percent versus the poorest fifth's 5 percent is a 9-fold gap, since 45 ÷ 5 = 9.

Now picture the chart you would draw. The horizontal axis runs from 0 to 100 percent of households, poorest first. The vertical axis runs from 0 to 100 percent of income. The line of perfect equality is a straight diagonal from the bottom-left corner to the top-right corner: if every fifth got a fifth, the curve would pass through 20, 40, 60, 80, and 100. Dataset S's Lorenz curve starts at the same bottom-left corner and ends at the same top-right corner, but in between it sags below the diagonal. At 60 percent of households it reaches only 30 percent of income. The curve bends upward more steeply toward the right, because the richer fifths add bigger slices.

In practice (your tools). In Desmos, open a table and enter the cumulative household shares 0, 0.2, 0.4, 0.6, 0.8, and 1 in the first column. Enter the cumulative income shares 0, 0.05, 0.15, 0.3, 0.55, and 1 in the second column, and turn on the lines that join the dots. Type y = x on a new line to draw the line of perfect equality. In Google Sheets, put the household shares in cells A2 to A7 and the income shares in B2 to B7, select both columns, and choose Insert, then Chart, then Line chart.

Misconception check. Students often plot each quintile's own share instead of the running total, which gives a line through 5, 10, 15, 25, and 45. That line never reaches 100 percent, so it is not a Lorenz curve. The cure: a Lorenz curve plots cumulative shares, and it ends at the top-right corner.

11. Compute the Gini coefficient

A curve is useful, but a single number makes comparisons easier. That number is named for Corrado Gini, who proposed it in 1912.

The Gini coefficient is a single number that summarizes how far a Lorenz curve falls below the line of perfect equality, ranging from 0 for perfect equality to 1 when one recipient or group receives all the income.

That definition follows the Census Bureau's own description of its Gini index. On a chart whose axes both run from 0 to 1, the triangle under the line of perfect equality has an area of 1⁄2 × 1 × 1 = 0.5. The Gini coefficient compares the gap between the diagonal and the Lorenz curve with that 0.5. A bigger gap means a bigger Gini.

You find the area under the Lorenz curve by cutting it into five strips, one per quintile. Each strip is a trapezoid 0.2 wide, with a left height and a right height read from the cumulative shares written as fractions.

Worked example — Dataset S's Gini coefficient. Work it strip by strip.

Each trapezoid's area. A trapezoid 0.2 wide has area 0.2 × (left + right) ÷ 2 = 0.1 × (left + right).
The five sums of heights. 0 + 0.05 = 0.05; 0.05 + 0.15 = 0.20; 0.15 + 0.30 = 0.45; 0.30 + 0.55 = 0.85; 0.55 + 1.00 = 1.55.
The area under the curve. 0.1 × (0.05 + 0.20 + 0.45 + 0.85 + 1.55) = 0.1 × 3.10 = 0.31.
The Gini coefficient. 1 − 2 × (area under the curve) = 1 − 2 × 0.31 = 1 − 0.62 = 0.38.
The same answer from the gap. The gap area is 0.5 − 0.31 = 0.19, and 0.19 ÷ 0.5 = 0.38.

In words: Dataset S's Lorenz curve falls well below the line of equality, and its Gini coefficient is 0.38 on a scale from 0 to 1. Someone comparing two regions would read a higher Gini as a more unequal distribution of income.

You can check the method against the benchmark. If every fifth received a fifth of income, the cumulative shares would be 20, 40, 60, 80, and 100 percent. The area would be 0.1 × (0.2 + 0.6 + 1.0 + 1.4 + 1.8) = 0.1 × 5.0 = 0.5, and the Gini coefficient would be 1 − 2 × 0.5 = 0. Perfect equality scores 0, as the definition says.

In practice (your tools). In the Google Sheets table from Section 10, type =(A3-A2)*(B2+B3)/2 in cell C3 and drag it down to C7 to get each trapezoid's area. Then type =SUM(C3:C7) in C8 for the area under the curve and =1-2*C8 in C9 for the Gini coefficient. With Dataset S the sheet shows 0.31 and 0.38.

One caveat: five quintiles give only five points, so the trapezoids cut a little below a smooth curve and the Gini is approximate. Official estimates, such as the Census Bureau's, are built from far more detailed data than five shares.

Misconception check. Students often read a Gini of 0 as perfect inequality, because a zero sounds like "nothing for anyone." It is the opposite. The cure: 0 means every household gets the same share, and 1 means one recipient or group gets all the income.

12. Weigh inequality as a named debate

The Lorenz curve and the Gini coefficient describe how income is shared. They do not say why it is shared that way, or what, if anything, a society should do about it. Those questions are contested, and this course presents them as a debate.

Start with the positive questions, the ones you can test with evidence.

How unequal are incomes, and how has that changed? Answering needs a real, dated source, such as the Census Bureau's published Gini index.
How much do differences in skills and training (Week 13's human capital), changes in technology and trade, and family background explain? Economists measure each of these and disagree about their relative size.
How do taxes and transfers change people's choices to work, save, and invest? This, too, is measured, and the estimates differ.

Then come the normative questions, which rest on values: how much equality a society should aim for, and whether equal opportunity or more equal outcomes matters more. Week 6 called this the question of equity, kept on the far side of the positive and normative line from efficiency.

Now the two broad cases, each at full strength.

The case for doing more to reduce inequality. Supporters argue that an extra dollar does more for a household with little than for one with plenty, so moving income toward lower-income households can ease real hardship. They point out that family background shapes the skills a child can build, so inequality of outcomes today can become inequality of opportunity tomorrow. Their normative claim is that a society should limit how far apart its members' lives can drift.
The case for caution about redistribution. Critics argue that taxes and transfers can weaken the rewards to working, saving, and starting firms, so redistribution can shrink the total that is shared. They point out that some inequality reflects choices, effort, and risk-taking that a market rewards. Their normative claim is that a fair process, with open rules and equal rights, matters more than an equal result.

Both sides accept the same positive tools: the Lorenz curve describes the distribution, and incentives shape behavior. They disagree about how large the effects are and how to weigh equity against efficiency. This course gives no verdict; the choice rests on evidence and on values.

Pause and reflect — choose one. Pick either prompt; they count the same.

(a) In two or three sentences, state the strongest point for doing more to reduce inequality, and label which part is positive and which is normative.
(b) In two or three sentences, state the strongest point for caution about redistribution, and label which part is positive and which is normative.

Misconception check. Students often think a Gini coefficient tells you whether a distribution is fair. It measures how unequal the shares are, a positive fact. The cure: whether 0.38 is too high, too low, or about right is a normative judgment that the number alone cannot settle.

13. Look back: when markets work and when they fail

This is the last teaching week, so step back and look at the whole term at once, everything you have built since Week 1. Every week built one model, and together they answer one question: when does a market allocate goods well, and when does it fail?

Markets allocate well when buyers and sellers face the full costs and benefits of their choices, when no one has the power to set the price alone, and when both sides can see what they are trading. Under those conditions the course's running market settled at 6 dollars and 30 units a day, with a total surplus of 90 dollars a day, the most that market can produce.

Each later week broke one of those conditions and measured the cost with the same surplus tools.

The term's market failures and policy cases, with the pinned illustrative losses from earlier weeks
What breaks	Where you met it	What it costs in the model
A binding price ceiling	Week 7, the ceiling at 4 dollars	A shortage of 40 and a deadweight loss of 40 dollars a day
A per-unit tax with no spillover	Week 7, the 2-dollar tax	20 units traded instead of 30, and a deadweight loss of 10 dollars a day
Market power	Week 11, the town's only ferry	8 hundred tickets instead of 16, and a deadweight loss of 32 hundred dollars a day
Rivals stuck in a dilemma	Week 12, the two coffee shops	A loss to the shops, 6 thousand dollars a month each instead of 10, while their customers gain from the low price
An external cost	Week 14, the smoke	30 units instead of the efficient 20, and a deadweight loss of 10 dollars a day
Hidden quality	Week 15, the used cars	Good cars that never trade, 1,000 dollars of surplus lost on each

Notice what the table tells you about policy. The same 2-dollar tax caused a deadweight loss of 10 in Week 7 and removed one of 10 in Week 14. A policy is not good or bad by itself; its effect depends on the market it lands in. That is the first reason to do the positive analysis before anything else.

Worked example — weighing a policy response. Use the three-step habit on any proposal, here a rule that every used-car dealer must offer a short warranty, the kind of promise Adaeze could not get in that parking lot.

Positive: what does the model predict? A warranty is a signal, so buyers can tell good cars from lemons, and good cars can sell between 9,000 and 10,000 dollars again.
Positive: who gains, who pays, and how much? Each good-car sale recovers 1,000 dollars of surplus; dealers bear the cost of the repairs they promise, and some of that cost may show up in prices.
Normative: is the trade-off worth it? Whether the recovered trades justify the cost to dealers and buyers rests on values, and you label it as a judgment.

In words: you can check the first two steps against evidence, and you cannot check the third, so an honest argument keeps them apart.

Misconception check. Students often take "the market fails" to mean "the government should take over." A market failure tells you the market alone misses the efficient outcome. The cure: whether a tax, a rule, a bargain, a signal, or no action does better is a separate question, part positive and part normative, and each response has costs of its own.

14. Check the classic wrong turns

Most mistakes this week come from swapping two ideas that sound alike or from reading a pattern as a verdict. A short routine you can run in your head catches them.

Step one: when is the hidden thing? A trait before the deal is adverse selection; an action after it is moral hazard.
Step two: who acts? The informed side signals; the uninformed side screens.
Step three: which end of the scale? A Gini of 0 is perfect equality and 1 is perfect inequality.
Step four: pattern or verdict? A bias is a predictable pattern, and a Gini is a measurement; neither one settles a normative question.

Worked example — checking a classmate's work. A classmate writes: "The lemons model proves used cars are bad. Insurance making drivers careless is adverse selection. Region A's Gini of 0 means one family gets everything."

The lemons claim. Half the cars in Dataset R are good; the model shows that good cars can be driven out when buyers cannot tell them apart.
The insurance claim. Taking less care after buying insurance is a hidden action after the deal, so it is moral hazard.
The Gini claim. A Gini of 0 means every household gets the same share; one family getting everything would be a Gini of 1.

In words: the corrected statement is "hidden quality can push good cars out, careless driving after insurance is moral hazard, and a Gini of 0 is perfect equality."

Misconception check. Students often think the behavioral patterns overturn the whole course. They do not. The cure: the supply-and-demand model still predicts well in many markets; behavioral economics adds the predictable cases where real choosers depart from it, and Akerlof's lemons add the cases where information is missing.

Key terms

Asymmetric information — A situation in which one side of a deal knows something relevant that the other side does not, such as a seller who knows a used car's history while the buyer gets one short test drive. (first used: 1. Name the gap: asymmetric information)
Adverse selection — The problem of hidden characteristics before a deal: when buyers cannot see quality, the sellers keenest to trade at an average price hold the worst goods, which can drive high-quality goods out of the market. (first used: 1. Name the gap: asymmetric information)
Moral hazard — The problem of hidden actions after a deal: once one side is protected, for example by insurance, it may take less care in ways the other side cannot observe. (first used: 1. Name the gap: asymmetric information)
Lemon — A used good, such as a car, with hidden defects that its seller knows about and a buyer cannot see before the sale. (first used: 2. Set up the used-car market)
Signaling — An action by the better-informed side of a deal that would cost too much to fake, such as offering a warranty, which reveals what that side knows about quality. (first used: 4. Rescue the trade: signaling and screening)
Screening — An action by the less-informed side of a deal to sort the other side by hidden quality or risk, such as paying for an independent inspection before buying. (first used: 4. Rescue the trade: signaling and screening)
Deductible — The part of each loss that an insured person pays before the insurer pays the rest, a common response to moral hazard because it keeps some cost of carelessness with the person who can prevent it. (first used: 5. Spot the hidden action: moral hazard)
Behavioral economics — The study of the predictable ways real people's choices depart from the fully careful, rational chooser assumed in the standard models. (first used: 6. Meet the real chooser: bounded rationality)
Bounded rationality — The idea that people make choices with limited time, attention, and information, and so rely on rules of thumb rather than weighing every cost and benefit. (first used: 6. Meet the real chooser: bounded rationality)
Loss aversion — The tendency for a loss to weigh more in a person's choices than a gain of the same size, a central element of Kahneman and Tversky's prospect theory. (first used: 7. Feel the sting: loss aversion)
Present bias — The tendency to weigh costs and benefits that arrive now much more heavily than the same costs and benefits later, so that plans to save or study keep being put off. (first used: 8. Spot present bias and anchoring)
Anchoring — The tendency for a first number a person sees, such as an original price on a tag, to pull later estimates and judgments toward it. (first used: 8. Spot present bias and anchoring)
Default — The option that takes effect when a person makes no active choice, such as automatic enrollment in a retirement plan that an employee may leave. (first used: 9. Set the default: the power of defaults)
Quintile — One of five equal groups formed by ranking households from the lowest income to the highest, so each group holds a fifth of all households. (first used: 10. Draw the Lorenz curve)
Cumulative share — The percent of all income received by one group of households together with every poorer group, the running total that a Lorenz curve plots. (first used: 10. Draw the Lorenz curve)
Lorenz curve — A graph of the cumulative share of income against the cumulative share of households, ranked from poorest to richest; the farther it sags below the line of perfect equality, the more unequal the distribution. (first used: 10. Draw the Lorenz curve)
Line of perfect equality — The straight diagonal on a Lorenz chart along which every fifth of households receives a fifth of income, the benchmark the Lorenz curve is compared with. (first used: 10. Draw the Lorenz curve)
Gini coefficient — A single number that summarizes how far a Lorenz curve falls below the line of perfect equality, ranging from 0 for perfect equality to 1 when one recipient or group receives all the income. (first used: 11. Compute the Gini coefficient)

Summary

Asymmetric information means one side of a deal knows more than the other.
Adverse selection is hidden traits before a deal; moral hazard is hidden actions after it.
In the lemons model, an average offer drives good cars out and only lemons trade.
The informed side signals, as with a warranty; the uninformed side screens, as with an inspection.
Deductibles and monitoring are responses to moral hazard.
Behavioral economics studies predictable departures from the careful chooser.
Loss aversion, present bias, and anchoring are patterns, not proof of foolishness.
Defaults change behavior because people tend to stay where they are put.
A Lorenz curve plots cumulative income shares against cumulative household shares.
The Gini coefficient runs from 0, perfect equality, to 1.
Inequality's causes and remedies are a debate over evidence and values; the course gives no verdict.
Markets allocate well with full costs, no price-setting power, and shared information; analyze any policy positively first.

Check your understanding

Work these before looking at the answers — they are ungraded, and the answers (with reasoning) are at the bottom of the page.

In a market for used laptops (an illustrative case), half the laptops are good and half have a hidden battery fault. A good laptop is worth 600 dollars to a buyer and 500 to its owner; a faulty one is worth 200 to a buyer and 100 to its owner. Find the uninformed buyer's offer, say which owners sell at that offer, and say what happens to the price once buyers catch on.
Label each case as adverse selection or moral hazard, and give your reason. (a) A gym that offers unlimited classes attracts mostly members who plan to come every day. (b) A delivery driver with a company-paid fuel card stops looking for the cheapest station.
In an illustrative town, the five household quintiles receive 10, 15, 20, 25, and 30 percent of income. Find the cumulative shares, the area under the Lorenz curve by trapezoids, and the Gini coefficient, and say what the Gini means.
Label each statement as positive or normative. (a) "Automatic enrollment raised participation in that firm's retirement plan." (b) "Employers ought to enroll every worker automatically." (c) "A higher tax on top incomes would reduce how much people at the top choose to work."

Answers (worked reasoning — read after attempting)

The offer is the average buyer value: (600 + 200) ÷ 2 = 800 ÷ 2 = 400 dollars. Owners of good laptops need 500, and 400 < 500, so they refuse; owners of faulty laptops need 100, and 400 > 100, so they sell. Once buyers learn that only faulty laptops are offered, they pay no more than 200 dollars, so faulty laptops trade between 100 and 200 and the good ones leave the market. Each good laptop that does not sell wastes 600 − 500 = 100 dollars of surplus.
(a) is adverse selection: how often a person plans to come is a hidden trait the gym cannot see before the membership is sold, and the members most eager to join are the most costly to serve. (b) is moral hazard: the driver's search effort is a hidden action that changes after the fuel card protects them from the cost.
The cumulative shares are 10; 10 + 15 = 25; 25 + 20 = 45; 45 + 25 = 70; and 70 + 30 = 100 percent. The sums of heights are 0 + 0.10 = 0.10; 0.10 + 0.25 = 0.35; 0.25 + 0.45 = 0.70; 0.45 + 0.70 = 1.15; and 0.70 + 1.00 = 1.70, which add to 4.00. The area under the curve is 0.1 × 4.00 = 0.40, and the Gini coefficient is 1 − 2 × 0.40 = 1 − 0.80 = 0.20. In words: this town's curve sits closer to the line of perfect equality than a curve with a higher Gini, so its incomes are less unequally shared, and whether that level is fair is a separate, normative question.
(a) is positive: it reports what happened and can be checked against the firm's records. (b) is normative: "ought to" signals a value judgment about what employers should do. (c) is positive: it is a prediction about behavior that evidence can test, even though economists disagree about how large the effect is.

Next up in this module: the Readings & Resources page, the lecture tutorial, the practice set, and the quiz — and next week the final exam asks you to use every model of the term.

Built with The Course Maker · samples.thecoursemaker.com