Chapter 1
When Markets Fail: The Hidden Rules of Who Gets What
Imagine a world where your kidney could save a life, but laws prevent you from donating it to a stranger. Or where getting into your dream school depends not on listing it as your first choice, but on understanding complex strategic games that privileged families master while others remain in the dark. This is the world Alvin Roth illuminates in his groundbreaking work on market design-a field that earned him the 2012 Nobel Prize in Economics. "Who Gets What and Why" has been hailed by Freakonomics author Steven Levitt as "a book that will change the way you think about markets," while Bill Gates included it in his "5 Books Worth Reading This Summer" list. The book's insights have influenced everything from school choice systems in major cities to lifesaving kidney exchange programs that have facilitated thousands of transplants. By revealing the hidden architecture behind markets that shape our lives, Roth shows us that markets aren't just places where money changes hands-they're sophisticated matching mechanisms where both sides must choose and be chosen.
Chapter 2
Markets Beyond Money: The Matching Revolution
Markets are far more diverse and pervasive than most people realize. While commodity markets like stock exchanges operate through prices, many crucial markets in our lives-from college admissions to organ transplants-function as matching markets where both sides must choose each other. These markets don't just need prices to work; they need sophisticated rules and systems that ensure participants can find each other (thickness), make decisions without overwhelming complexity (managing congestion), and feel safe revealing their true preferences.
The transformation of agricultural products illustrates this evolution perfectly. Before the Chicago Board of Trade established wheat grading standards in 1848, buyers had to personally inspect each farmer's crop-a matching market where specific buyers chose specific sellers. With standardization, wheat became a commodity that could be purchased without inspection, making the market "thick" by allowing any buyer to transact with any seller. Similarly, Ethiopian coffee underwent this transformation in 2008 when the Ethiopia Commodity Exchange introduced professional tasters who grade coffee through blind testing. This standardization improved quality as farmers began selectively harvesting only ripe red cherries to earn higher grades and prices.
Even our digital experiences reflect market design principles. When you search on Google, you're participating in an auction marketplace where advertisers bid for your attention based on your search terms. The highest bidder's ad appears first, but they pay only the second-highest bid price-a mechanism designed to encourage honest valuation. These invisible auctions happen in milliseconds, creating a marketplace where even niche products can profitably advertise to pre-qualified prospects.
The distinction between matching markets and commodity markets isn't just academic-it fundamentally changes how we must approach market failures. When matching markets fail, the solution isn't simply adjusting prices or increasing supply; it requires redesigning the rules and mechanisms by which participants find and choose each other.
Chapter 3
Lifesaving Exchanges: Kidney Transplantation Reimagined
What do you do when someone needs a kidney transplant but their willing donor isn't compatible? Until recently, these patient-donor pairs faced a devastating dead end. Dr. Michael Rees, a transplant surgeon, grew increasingly frustrated watching patients die while willing donors sat helplessly by. His initial attempts to manually match incompatible pairs by comparing charts at his kitchen table failed, but he recognized the potential of a larger, systematic approach.
The theoretical foundation for kidney exchange had actually been laid decades earlier. In 1974, game theorists Lloyd Shapley and Herb Scarf published a model for trading indivisible goods without money-a perfect framework for kidney exchange. Their "top trading cycles" showed how cyclical trades could be arranged where each participant received their most preferred available option. Alvin Roth extended this work, proving that such systems could safely allow participants to reveal their true preferences without strategic manipulation.
When Roth and his colleagues shared their theoretical work with surgeons, they encountered practical limitations. Dr. Frank Delmonico pointed out that exchanges initially needed to be limited to two pairs at a time due to the logistical challenge of coordinating simultaneous surgeries. This led to new algorithms that accounted for real-world constraints while still making it safe for patients and surgeons to share necessary information.
A breakthrough came with the recognition that non-directed "altruistic" donors could initiate chains of transplants. In 2010, John Robertson, a 62-year-old semi-retired bookstore owner, became such a donor. His kidney went to Jack Burns, whose wife Adele then donated her kidney to another patient. This chain approach doubled the transplants from a single donation. Eventually, the team implemented nonsimultaneous chains, where surgeries could happen over time rather than all at once. The first such chain, started by donor Matt Jones in 2007, eventually resulted in sixteen transplants across multiple states over several years.
The success of kidney exchange reveals a profound truth about market design: sometimes the most valuable innovations aren't technological breakthroughs but new ways of organizing existing resources. By creating systems that allow people to safely reveal their needs and capabilities, market designers can unlock tremendous value that would otherwise remain trapped by incompatibility.
Chapter 4
When Markets Rush: The Problem of Timing
Markets fail when timing goes wrong, often manifesting as a "rush" to be first. This phenomenon appears across diverse contexts-from fraternity recruitment to professional hiring to marriage markets. The term "rush" in Greek organization recruitment originated in the late 1800s when these organizations competed to recruit students earlier and earlier. Similarly, "Sooners" in Oklahoma got their nickname by entering land before the official start of land runs.
College football bowl games perfectly illustrate timing problems. For years, bowls recruited teams too early in the season, before important games were played. In 1990, despite NCAA rules prohibiting early arrangements, Notre Dame (ranked #1) and Colorado (#2) agreed to play in the Orange Bowl thirteen days before the official selection date. When Notre Dame subsequently lost and finished ranked #5, the championship became contested. The market only improved when conferences formed the Bowl Championship Series, creating a system where top teams could be matched after all regular season games concluded.
Professional markets suffer similar timing pathologies. Investment banks see their new hires poached by private equity firms before they've even started work. Law firms hire summer associates two years before graduation, committing to employment without knowing their future needs. This "unraveling" proved disastrous during the 2008 recession, when firms had to rescind thousands of offers made to students they'd recruited long before.
Federal appellate clerkships exemplify prestigious positions where unraveling occurs despite seemingly counterintuitive conditions. While there are few judges compared to law students, judges compete for the even smaller pool of top students. Less prestigious judges make early offers to secure talent before more prestigious circuits can. Students face difficult strategic decisions-accept an immediate offer from a good circuit or risk waiting for a potentially better one.
Marriage markets demonstrate similar dynamics across cultures and history. When markets are "thin" (few available partners), people make earlier commitments than when markets are "thick" (many options). In some polygamous societies where women are scarce, marriages are arranged extremely early-even for infants. The Arunta aboriginal people took this to an extreme, arranging marriages between an infant boy and the future daughter of an infant girl-effectively arranging marriages more than a generation in advance.
The gastroenterology fellowship market suffered similar unraveling until researchers designed a solution: rather than trying to enforce a hiring date (which directors didn't trust each other to follow), they empowered applicants to change their minds about early acceptances. This removed the incentive for early offers, as program directors couldn't "capture" desirable candidates prematurely.
Chapter 5
Speed Kills: When Markets Move Too Fast
While unraveling makes markets operate too early, another speed-related problem occurs when markets move too fast. In thick markets, participants want quick evaluation of many potential transactions, but excessive speed can create serious problems.
Financial markets exemplify this issue. High-speed trading harms market function by forcing liquidity providers to quote bigger spreads to protect themselves from being "sniped" by traders using superfast connections. These wider spreads get passed on to ordinary investors as costs. Very high-speed trading also contributes to market instability, as seen in the 2010 "flash crash" when prices plunged and recovered in just four minutes due to computer algorithms trading at speeds beyond human supervision.
This speed-based competition resembles insider trading in how it disadvantages ordinary investors. Eric Budish and colleagues proposed a simple solution: run markets once per second rather than continuously, so trades go to those offering the best prices rather than the fastest traders, restoring price competition and eliminating the wasteful arms race for speed.
While millisecond-level speed creates problems in modern financial markets, appropriate speed increases can benefit markets. In the nineteenth century, the transatlantic telegraph cable dramatically improved the cotton market by reducing information transmission time between America and England from weeks to a day. Before the cable, price information took ten days to cross the ocean, making it difficult to match supply with demand. After the cable's completion, cotton shipments better matched market fluctuations and prices became less volatile.
In Japan's university graduate hiring market, companies found ways to circumvent timing agreements dating back to 1953. Rather than making official early offers (which would be embarrassing), companies created "slow-motion exploding offers" by scheduling mandatory events on days when civil service exams were held. Students who didn't attend these events wouldn't receive promised job offers later, effectively forcing them to choose without seeing all options.
A 1984 survey revealed 88.4% of major Japanese companies thought hiring agreements should continue, while 87.7% admitted not following them. This demonstrates that timing rules alone aren't sufficient to create thick markets where people can consider multiple opportunities.
Chapter 6
Congestion: The Thickness Dilemma
Markets follow a "Goldilocks principle" regarding speed: transactions must be neither too fast nor too slow. Congestion occurs when markets become thick but participants don't have adequate time to consider multiple offers. Even internet markets face this challenge, as human decision-making creates bottlenecks despite digital speed.
Airbnb illustrates this problem. When it launched in 2008, hosts might post rooms in the morning and check for interest in the evening. Potential guests could wait a full day to learn if a room was available, then have to start over if rejected. This worked when the market was small and budget-conscious, but as Airbnb grew, users increasingly needed multiple attempts to secure reservations. Meanwhile, hotel chains offered the advantage of instant confirmation through centralized booking systems.
In 2003, New York City's Department of Education sought help redesigning their school assignment system. With 90,000 ninth graders annually choosing from hundreds of high schools, the market was thick but severely congested. The paper-based system required students to list five preferred schools, with schools receiving and reviewing these applications. After three rounds of offers, acceptances, and waiting lists, approximately 30,000 students remained unassigned until August. The process was not only slow but also unsafe-savvy parents circumvented the system through direct appeals to principals, who often held back seats from the official process.
Thick markets need speed, but human decision-making creates bottlenecks. In housing markets, buyers make offers with "earnest money" and short timeframes. While sellers consider these offers, buyers can't pursue other properties, and desirable homes may sell in the meantime. Job markets function similarly-once a company makes an offer, it must wait for the candidate's response while other candidates may accept positions elsewhere.
The solution to New York's school choice congestion was letting people submit all preferences at once, then using a computer to process decisions quickly. This approach works in other matching markets too, with computerized clearinghouses that process preferences people decide on before the computer runs. The key is making it safe for people to state their preferences honestly.
Chapter 7
Trust, Safety, and Simplicity: Market Foundations
Making markets safe is one of the oldest problems of market design. Medieval kings provided safe passage to markets, protecting commerce from highwaymen. Modern markets still face physical dangers, particularly illegal ones, but also more common risks like not receiving goods as expected or having credit card information stolen. Legal marketplaces offer recourse against dishonest merchants, while reputations help customers identify trustworthy businesses.
Internet marketplaces have focused on secure payment methods, transaction insurance, and feedback systems to build trustworthiness. eBay pioneered reputation systems where buyers and sellers could rate each other. Initially, feedback became overwhelmingly positive due to reciprocity-users exchanged positive ratings following the "scratch my back" principle. Working with economists, eBay redesigned their system to allow anonymous detailed feedback, which improved information quality.
While markets need information to function well, too much information sharing can compromise privacy and make markets unsafe. On eBay, bidders engage in "sniping"-placing bids in the final seconds of auctions to avoid revealing their true valuation and triggering bidding wars. Despite the risks of forgetting to bid or having bids arrive too late, most snipers feel this strategy is safer than revealing their maximum willingness to pay early.
Unlike New York's congestion crisis, Boston's school assignment system appeared to run smoothly on the surface while hiding a fundamental flaw: it wasn't safe for families to express their true preferences. Boston used an "immediate acceptance" algorithm that filled schools with students who listed them first, making it strategically dangerous to list a popular school as first choice unless you had priority there. Parents who listed their genuine top choice risked getting none of their preferred schools if that choice was competitive.
Consequently, about 80% of children received their "first choice" schools-but many of these represented safe, strategic choices rather than true preferences. This forced families into complex strategic decisions, creating what one parent called "the alienation of not making your first choice your first choice."
Boston Public Schools senior staff realized there might be a problem with their system after meeting with economists. The economists' analysis revealed families needed to strategically misrepresent preferences to succeed in the system, with different consequences for different families. About 20% of parents listed popular schools as second choices that were impossible to get except as first choices, resulting in unassigned children. Some affluent parent groups gathered playground "intelligence" about applications and sibling priorities to estimate admission chances.
Chapter 8
The Match: Designing Markets That Work
Market design solutions are sometimes invented, sometimes discovered, and often both-like antibiotics, which were discovered naturally but required invention to become practical medicine. The market for new doctors provides an instructive example of addressing common matching market failures.
Since 1900, American medical graduates have required supervised internships or residencies before practicing independently. By 1940, competition among hospitals had caused "unraveling," with residents hired nearly two years before graduation-too early to accurately assess students' abilities or preferences. Medical schools finally controlled timing in 1945 by embargoing student information before a specified date, but this created a new problem: congestion. With simultaneous offers, hospitals made exploding offers requiring immediate responses, creating a chaotic, rushed market with missed and broken agreements.
In 1952, doctors revolutionized the process by creating a centralized clearinghouse where students and programs submitted rank-ordered preferences after interviews. An initial flawed algorithm was challenged by Harvard medical student Hardy Hendren, who organized opposition through the National Student Internship Matching Committee. The revised "Boston Pool Plan" succeeded because it produced "stable" outcomes-matches where no unmatched student-program pair would prefer each other over their assigned matches.
The medical Match worked smoothly for decades until women began entering medical school in significant numbers in the 1970s. This created a new challenge: married couples needed two residencies near each other. The original algorithm couldn't accommodate this need, causing couples to decline Match offers and arrange jobs directly with hospitals, which destabilized the entire system.
In 1995, Roth partnered with Elliott Peranson to develop the Roth-Peranson algorithm-a hybrid approach that starts with a deferred acceptance algorithm and then systematically fixes blocking pairs involving couples. Despite theoretical concerns that stable matchings might not always exist with couples in the market, they found that in practice, stable outcomes were almost always possible.
When redesigning the Match, Roth discovered that centralized marketplaces differ fundamentally from central planning. Some medical administrators hoped he could solve the problem of rural hospitals struggling to attract residents. However, Roth's earlier mathematical work (the Rural Hospitals Theorem) had already proven this impossible: hospitals that don't fill all positions at one stable outcome get exactly the same doctors at every stable outcome.
Following their success in New York and Boston, Roth and his colleagues have helped design school choice mechanisms for Denver, New Orleans, and Washington D.C., with more cities in progress. Their nonprofit Institute for Innovation in Public School Choice, led by Neil Dorosin, has become instrumental in spreading these market design principles. Similar efforts have begun in Japan, Belgium, England, and China.
Chapter 9
Repugnance and Market Boundaries
Some transactions become impossible not because of practical failures but because society deems them repugnant. In California, it's illegal to sell horsemeat for human consumption-not for food safety reasons, but because 60% of voters in a 1998 referendum found the practice objectionable. This illustrates what Roth calls "repugnant transactions"-exchanges some people want to engage in that others object to, even when they themselves experience no direct harm.
Repugnance varies by location and culture. Transactions can be repugnant but legal when insufficient people oppose them or when prohibition proves unenforceable, as demonstrated by America's failed Prohibition experiment (1920-1933), which fueled organized crime. Conversely, some transactions are "protected"-like small farming-where many support others' rights to engage in them without participating themselves.
Repugnance can change rapidly over time, as with same-sex marriage's swift legalization across America, while other repugnances like those against polygamy and slavery have strengthened. Interest-bearing loans, once forbidden to Christians, became accepted as attitudes shifted from Shakespeare's cautionary "neither a borrower nor a lender be" to Benjamin Franklin's view of responsible credit as a virtue.
Some transactions become repugnant only when money enters the equation. Gifts and in-kind exchanges that are celebrated when freely given-like loans without interest, child adoption, or expressions of love-often face disapproval when commercialized. This explains why dinner guests bring wine rather than offering cash payment.
The kidney shortage presents a classic design challenge: while Iran permits a legal kidney market, most countries forbid cash payments despite the 100,000 Americans waiting for transplants with only 17,000 performed annually. This gap illustrates how repugnance can block transactions even when lives are at stake, forcing designers to find creative alternatives that address ethical concerns while increasing supply.
If kidney sales were permitted, careful market design could address common objections: a single government buyer could ensure fair allocation rather than favoring the wealthy; a one-year cooling-off period with thorough screening could prevent exploitation; and generous compensation (affordable given Medicare's dialysis savings) could ensure sellers aren't exploited.
Given the widespread legal prohibition of kidney sales, we might instead focus on removing donation disincentives through covering expenses and lost wages, as Israel has done with its forty-day wage compensation program. Kidney exchange offers another promising path forward without triggering repugnance concerns. While non-directed living donors currently start chains yielding about five transplants each, deceased donors could be incorporated into similar chains, potentially doubling transplantation rates.
Chapter 10
Markets as Human Artifacts: Design Matters
Markets are human artifacts that can be designed to deliver benefits while minimizing harms, not simply binary yes-or-no propositions. The concept of "free markets" doesn't mean markets without rules. As Friedrich Hayek noted in The Road to Serfdom, there's "all the difference between deliberately creating a system within which competition will work as beneficially as possible and passively accepting institutions as they are." Markets need effective rules to work freely-like a wheel needs an axle and bearings to rotate smoothly.
Restaurants demonstrate key market principles in action. Different restaurant types handle congestion differently: fine dining establishments use reservations to manage kitchen congestion (waiting happens at the table), mid-range restaurants use waiting lists to handle dining room congestion (waiting happens before seating), while fast food places minimize transaction time with assembly-line cooking and simplified ordering. These operational differences extend to visible markers like tablecloth choices-white linen, plastic, or none-perfectly aligned with each restaurant's congestion management approach.
Bad market designs can persist despite their flaws, partly because coordinating change requires overcoming stakeholders invested in the status quo. The U.S. healthcare payment system exemplifies poor design-third-party payers lack incentives to reduce costs or focus on prevention, making it easier to finance expensive treatments than preventative care. California's water rights system allocates water without regard to its highest value use, preventing efficient transfers during droughts-a market failure that persists despite obvious inefficiency.
Computers have revolutionized markets, making them ubiquitous and fast while enabling "smart markets" that rely on complex calculations. Kidney exchanges and package bidding auctions would be impossible without computational power to match patient-donor pairs or determine optimal spectrum license allocations. Google's advertising auctions demonstrate how computational speed isn't merely convenient-it's essential for creating entirely new market possibilities.
Markets are like languages-ancient human inventions that help us organize ourselves, cooperate, coordinate and compete. Both constantly adapt to changing circumstances. Markets come in many varieties, from impersonal commodity markets to deeply personal matching markets that shape pivotal life moments.
Economists traditionally studied markets as natural phenomena beyond our control, but we now understand we can intervene, redesign, fix, and create markets. This shift toward economists as engineers resembles transformations in farming and medicine. Early farmers simply grew what they found; modern agriculture benefits from centuries of deliberate breeding and genetic engineering. Similarly, medicine evolved from merely predicting outcomes to actively intervening in diseases.
Markets, like languages, can be difficult to redesign because they're collective tools, but when opportunities arise to fix broken markets or design new ones, we should approach them with careful study and humility, recognizing that markets are human artifacts, not natural phenomena. By understanding the hidden rules that determine who gets what-and why-we can create more efficient, fair, and functional systems that better serve human needs.
