Skip to content

Money & Work

11 min left
BOOK SUMMARY · ECONOMICS

Freakonomics Summary: The Key Ideas, What Holds Up, and How to Use Them

Levitt and Dubner's tour of hidden incentives, and which of its studies still hold up.

Hacks Vitae artwork (AI-assisted)
The answer

Freakonomics uses data to look for the incentives hidden behind everyday behaviour, from cheating teachers and sumo wrestlers to estate agents and a street gang.

It suits readers wanting an entertaining introduction to how economists use data. Most studies hold up within their scope; a few rest on one city or one gang, and the abortion–crime link is still argued.

Why it mattersKnowing which studies held up tells you which of the book's lessons travel beyond one city or one gang.

Key facts
  1. IncentivesEconomic, social, moral; one day-care fine made lateness rise.
  2. CheatingData patterns pointed to cheating teachers and sumo wrestlers.
  3. InformationEstate agents' own homes sold for 3.7% more, in one study.
  4. Gang financesOne gang's street dealers appear to have earned below minimum wage, most years.
  5. Distant causesThe abortion–crime link is still argued among economists.
  6. Who parents areChicago lottery winners showed little systematic academic benefit.

By Hacks Vitae Published Updated Last checked October 2026 11 min read 28 sources ↓

Start hereFining late parents at Israeli day-care centres made lateness go up. A later survey-based replication found fines made respondents less likely to be late. So which of the book's stories travel?

About the book

Freakonomics is a best-selling book about the hidden incentives behind everyday behaviour. It is by Steven D. Levitt, a University of Chicago economist who received the John Bates Clark Medal in 2003, and Stephen J. Dubner, an award-winning author and journalist.

William Morrow published it in 2005, and a revised and expanded edition followed in 2006. The authors’ website says it went on to sell more than 4 million copies in 35 languages.

Each chapter takes an odd question, such as what schoolteachers and sumo wrestlers have in common. It answers with a story built on research, much of it Levitt’s own papers.

Incentives explain a great deal

The book calls incentives “the cornerstone of modern life”. It sorts them into three flavours: economic, social and moral.

An early example is a group of Israeli day-care centres that began fining parents who came late, and saw lateness go up rather than down. The researchers argued that a fine can change the information people have about what lateness means.

Does it hold up? Partly. In the original field study, the number of late parents increased significantly, and it did not fall again after the fine was removed.

A 2020 replication used online surveys on MTurk and did not reproduce the result. There, fines made respondents less likely to be late, and the effect disappeared once the fine was removed. Its authors note that a survey is not a field experiment, and that more research is needed to know when a fine works as a price.

Try it: after a new rule or penalty, look at what people did next rather than assuming it worked.

Where there are incentives, there is cheating

Chicago schoolteachers facing test-based incentives and sumo wrestlers one win short of a winning record both had reasons to cheat. Levitt’s papers looked for patterns in the data that honest behaviour would rarely produce, and found them.

Does it hold up? Yes, within the studies’ scope. Jacob and Levitt built an algorithm that combined unexpected jumps in test scores with suspicious patterns of answers in a classroom. Using Chicago Public Schools data, they estimated serious cheating by teachers or administrators in at least 4–5% of elementary classrooms each year.

The frequency of cheating responded strongly to relatively minor changes in incentives.

Duggan and Levitt found that wrestlers on the margin of a winning record won a disproportionate share of their matches. Extra effort could not explain it.

Wrestlers who won on the bubble also lost more often than expected the next time they met the same opponent. That suggests part of the payment was a favour returned.

In February 2011 the Japan Sumo Association cancelled its March tournament after police found emails listing 13 wrestlers that indicated some may have been involved in fixing matches.

Try it: Jacob and Levitt concluded that incentive systems, especially those with bright-line rules, often induce distortions such as cheating. So look first at what happens just around a target or bonus line.

Information is power

The book argues that experts, from estate agents to criminologists, use their informational advantage over the people who rely on them. Its example is a study comparing agents’ own home sales with their clients’.

Because an agent receives only a small share of any extra price, the agent has a reason to sell a client’s house too cheaply and too quickly.

Does it hold up? Yes, in the study the book draws on. Levitt and Syverson compared sales where agents were hired by others with sales of agents’ own homes.

Agents’ homes sold for 3.7% more and stayed on the market 9.5 days longer, controlling for observable features of the houses. The book rounds these to an extra 3-plus percent and ten days.

Another study, of condominiums, also found a price premium for agents’ own homes. It found that agent-owned condominiums stayed on the market longer, which the authors contrast with earlier results for single-family houses in the same market.

The Klan chapter applies the same lens to secrets. It tells how Stetson Kennedy upended the Klan’s information asymmetry in the 1940s, retelling Kennedy’s own account in his book The Klan Unmasked.

Does that story hold up? Partly. Kennedy did campaign against the Klan. He shared the names of prominent Klan members with journalists, aided authorities in Klan prosecutions and contributed to a Superman radio episode ridiculing the Klan.

Dubner and Levitt later examined archive documents. In January 2006 they wrote that many of the adventures Kennedy described as his own seem to have been based on the efforts of a different man. That man was a Klan informant named John Brown, and the authors said they were amending the chapter for later printings.

The conventional wisdom is often wrong

The authors write that “the conventional wisdom is often wrong”, and the gang chapter is their clearest case. One gang’s own financial records, analysed by Levitt and Sudhir Venkatesh, suggested that street-level drug dealing is less lucrative than is generally thought. The money was concentrated at the top.

That is the point of the chapter title, “Why Do Drug Dealers Still Live with Their Moms?”

Does it hold up? Partly: it is true for the one gang studied. Levitt and Venkatesh analysed the gang’s own monthly financial records over four years. The average wage in the gang rose from roughly $6 to $11 an hour, but pay was extremely skewed.

The leaders earned far more than they could in legitimate jobs. Street-level dealers appear to have earned less than the minimum wage for most of the period. That was despite an annual violent death rate in the sample of 0.07, or seven in a hundred.

One gang over four years is a case study, not a general rate.

Big effects can have distant causes

The crime chapter argues that the legalisation of abortion in the US in the early 1970s helped cause the fall in crime in the 1990s. Its reasoning, in one line: “Legalized abortion led to less unwantedness; unwantedness leads to high crime; legalized abortion, therefore, led to less crime.”

Does it hold up? It is still argued between credible researchers, with evidence on both sides.

Donohue and Levitt reported in 2001 that crime began to fall roughly 18 years after abortion was legalised. The five states which allowed it in 1970 saw declines earlier. They reported that legalised abortion appears to account for as much as 50 percent of the drop.

Foote and Goetz found a coding mistake in one table. They reported much weaker results after correcting it and measuring crime per person.

Donohue and Levitt acknowledged the omission. They said fixing it reduced the size of the estimates but not the conclusion. They argued that a better measure of abortion made the evidence as strong or stronger.

Critics argue that comparing states cannot separate abortion from everything else that differed between them. Foote put it this way in 2019: “there are a lot of reasons New York crime might decline relative to Utah crime.”

Using age-specific crime rates, Joyce found no meaningful association. Reviewing the evidence in 2009, Joyce wrote that criminologists largely dismiss the association, while Donohue and Levitt have responded to each challenge with more data.

In 2019 Donohue and Levitt tested their 2001 prediction on seventeen newer years of data and reported strong support.

A study of Romania’s 1966 abortion ban and 1989 legalisation found effects on the number of crimes but not on crime rates. It put them down to cohort size rather than unwantedness.

Independent tests are split:

Our reading: the size of the effect is unsettled rather than established.

Whether legal abortion is right or wrong is a moral question; none of this research answers it, and this page takes no side on it.

Who parents are matters more than what they do

The parenting chapter concludes that “it isn’t so much a matter of what you do as a parent; it’s who you are.” One strand of its evidence is Chicago’s high-school lotteries. Students who won a place at a sought-after school did not do systematically better on traditional academic measures than similar students who lost.

Does it hold up? Yes, for Chicago’s lotteries. Because admission to popular Chicago public high schools was decided by lottery, Cullen, Jacob and Levitt could compare winners with similar students who lost.

Winners attended schools that were better on several measures. Yet the authors found little evidence of any systematic benefit across a wide variety of traditional academic measures. They did find improvements in some non-traditional outcomes, such as self-reported disciplinary incidents and arrest rates.

A lottery study in Charlotte-Mecklenburg, North Carolina, later found a significant overall increase in college attainment among winners who attended their first-choice school. So the Chicago result is not a general rule about school choice.

Try it: students who apply for a popular school are not a typical group, which is why the Chicago researchers used lotteries. Before crediting a choice with a result, ask who made the choice.

What holds up, and what the book overstates

What the sources show for each idea

01 Some Chicago teachers cheated on their pupils' standardised tests Holds up One study estimated serious cheating in at least 4–5% of elementary classrooms a year. Evidence ↓
02 Sumo wrestlers threw matches when an opponent badly needed the win Holds up Strong statistical signs; in 2011 a tournament was cancelled after emails pointed to possible fixing. Evidence ↓
03 Most street-level drug dealers earn less than the minimum wage Partly True for the one gang studied over four years; a case study, not a general rate. Evidence ↓
04 A distinctively Black first name does not, by itself, hold a person back Partly No causal effect found on life outcomes, but résumé experiments find fewer callbacks. Evidence ↓
05 Winning a place at a sought-after Chicago high school did not improve students' academic results Holds up Little systematic academic benefit in Chicago's lotteries; a Charlotte study found gains in college attainment. Evidence ↓
06 Estate agents sell their own homes for more than their clients' homes Holds up Agents' own homes sold for 3.7% more in the study the book draws on. Evidence ↓
07 Fining parents for picking their children up late made lateness go up Partly Lateness rose in the original field study; a 2020 survey replication found the opposite. Evidence ↓
08 Legalised abortion explains up to half of the 1990s fall in US crimeCredited to Donohue and Levitt Still argued Independent studies point both ways; the size of the effect is unsettled. Evidence ↓
09 Stetson Kennedy went undercover inside the Ku Klux Klan himself, as the book tells it Partly He did campaign against the Klan; the authors later said many adventures seem another man's. Evidence ↓
6 of 9 shown · a mix of every verdict

Most of the book’s studies hold up within their own scope, and the main ones were published in peer-reviewed economics journals. A few stories are weaker than the book makes them sound, mostly because they rest on one study, one city or one gang. The abortion–crime link is the one real dispute.

First names. Levitt also studied distinctively Black first names with Roland Fryer. Fryer and Levitt found that such names became a strong signal of socio-economic status among Black children born in recent decades. They found no negative causal impact of the name itself on life outcomes.

Job-application experiments measure something else: an employer’s first response. In a study of help-wanted ads in Boston and Chicago, résumés with white-sounding names received 50 percent more callbacks.

A later experiment with more than 83,000 applications to 108 of the largest US employers found that distinctively Black names lowered the chance of contact by 2.1 percentage points. Fryer and Levitt argued that their finding and the résumé studies can be reconciled.

Who it’s for, and who can skip it

It suits readers who want an entertaining introduction to how economists use data to answer everyday questions. Readers looking for a practical guide or for up-to-date evidence will need to look beyond it.

The studies date from the late 1990s and early 2000s, and the debates about several of them continued long after publication.

Quick answers

What is Freakonomics about?

Levitt and Dubner use data to look for the incentives hidden behind everyday behaviour, from schoolteachers and sumo wrestlers to estate agents and a street gang.

Is the Freakonomics abortion and crime theory true?

It is still argued. Donohue and Levitt estimated that legalised abortion accounted for as much as half of the 1990s fall in US crime, and their 2019 update reported support for their own prediction. Critics found a coding error in one table, and independent studies point both ways. Whether abortion is right or wrong is a moral question this research does not answer.

Did sumo wrestlers really fix matches?

Duggan and Levitt found statistical signs of it. In February 2011 the Japan Sumo Association cancelled its March tournament after police found emails that indicated some wrestlers may have been involved in fixing matches.

Do drug dealers really live with their moms?

In the one gang Levitt and Sudhir Venkatesh studied, street-level dealers appear to have earned less than the minimum wage for most of the four years. One gang is a case study, not a general rate.

Use this

The lesson

Look for the reward behind a behaviour, then check the data before you trust the story.

The rule

What is each person here rewarded for, and would the pattern look the same if they weren't?

Quick quiz

Three questions, thirty seconds

Nothing is scored or saved. Recalling it once now is what makes it last.

1. In the Israeli day-care study, what happened after a fine for late pick-ups was introduced?

Answer: Lateness went up.Read it again ↑

2. In Levitt and Syverson's study, how did estate agents' own homes sell compared with other homes?

Answer: For 3.7% more.Read it again ↑

3. What did Chicago students who won a school-place lottery gain on traditional academic measures?

Answer: Little or no systematic benefit.Read it again ↑

How we checked this · Last checked October 2026
  • What we read of the book: passages of the 2005 and 2006 printings, searched through Open Library's search-inside tool, plus the publisher's and the authors' pages and the authors' January 2006 web page on the Kennedy story.
  • For each study we read the published abstract, not the full text. We read the France 24, NPR and Journalist's Resource articles (the last updated in 2022).
  • The 2020 day-care replication was a survey, not a field experiment; its authors say so, and the page reports both results.
  • On abortion and crime we report each side's published findings. We take no position on the moral question, which none of the research answers.

How we check every article →

Sources

Primary written at the time, by a witness, or the original text Scholarly edited or peer-reviewed work Popular press, blogs and general websites