The American Bar Association released its annual bar passage data on March 12, 2026, and the legal education establishment celebrated. Pass rates were up. Elite schools were posting near-perfect numbers. Stanford Law School's graduating class achieved a first-time bar passage rate of 99.43% — one student out of 176 failed. Yale came in at 98.54%. Harvard at 97.90%. The press releases wrote themselves. The admissions offices updated their websites.

Nobody mentioned the six schools at the other end of the list.

Six ABA-accredited law schools — institutions licensed to charge $150,000 to $250,000 for a three-year legal education — failed to meet the ABA's own minimum accreditation standard for bar passage in 2025. The standard is 75%: three out of four graduates must pass the bar exam within two years of graduation. It is the floor. It is the lowest possible bar. And six schools couldn't clear it.

One week later, Reuters published a story about a new academic study from Yale and Vassar — breathlessly headlined "Does a Law Degree Pay Off? New Study Says Yes" — that the legal education industry immediately held up as vindication. Parents can stop worrying. The investment makes sense. Law school is worth it.

The study's underlying data runs from the early 1990s to 2018.

That is not a footnote. That is the entire story.

The Two-Tier Reality the ABA's Numbers Reveal

When the legal education establishment discusses bar pass rates, it focuses on the top. Stanford at 99.43%. Yale at 98.54%. Duke at 98.23%. Harvard and SMU rounding out the top five. These are the numbers that make it into admissions brochures and alumni newsletters and the Reuters headline. These are the numbers that prospective law students read when they are trying to decide whether to apply.

What the brochures don't mention: there are 198 ABA-accredited law schools in the United States. Stanford, Yale, Duke, Harvard, and SMU are five of them. The other 193 schools are not Stanford, Yale, Duke, Harvard, or SMU. Most of their graduates are not walking into the same job market that elite school graduates enter. Most of them are not going to BigLaw. Most of them are not going to federal clerkships. Most of them are going into the general legal job market — the regional firms, the solo practices, the public defender offices, the government agencies, the positions that pay $70,000 to $95,000 per year in mid-sized markets.

And six of those 198 schools cannot even get three-quarters of their graduates past the bar exam within two years of graduation.

The ABA's accreditation standard for bar passage — 75% ultimate bar passage within two years — exists precisely because the bar exam is the threshold credential for practicing law. An attorney who cannot pass the bar cannot work as an attorney. A law school whose graduates cannot pass the bar at a 75% rate is, by the ABA's own definition, failing to produce attorneys. The degree it is selling — for $150,000 to $250,000, across three years of a young person's professional life — is not reliably delivering what it promises.

Six schools are below that floor. More are hovering just above it. And every one of them is still fully accredited, still accepting new students, still cashing tuition checks from young people who believe that a J.D. from an ABA-accredited institution is a ticket to a legal career.

The Mathematics of Bar Failure Nobody Advertises

In 2025, 66,174 people sat for bar examinations in the United States. Of those, 38,485 passed — a 58% overall pass rate when repeat takers are included. The 84% first-time pass rate that the legal establishment celebrates obscures this number entirely. It counts only people taking the bar for the first time. It excludes the significant population of repeat takers — graduates who failed once, twice, three times — who bring the overall pass rate down to 58%.

Think about what that means. More than one in four people who sat for the bar exam in 2025 failed. Many of them had already failed before. Many of them are carrying six-figure debt from a law school education that did not prepare them to clear the profession's most basic threshold requirement.

The ABA's ultimate bar pass rate — 92.15% to 92.25% for the class of 2023, measured over two years — sounds better. But it means that approximately 7-8% of law school graduates who attempt the bar cannot pass it within two years of graduation, despite having invested three years and $100,000 to $300,000 in their legal education. At a school charging $250,000 for a three-year degree, 7-8% of graduates are paying full price for a credential they cannot convert into a license to practice.

At the six schools below the 75% threshold, the picture is worse. A graduate of one of those schools has a coin-flip probability of passing the bar within two years. They are paying law school prices for bar exam odds that would be embarrassing at any LSAT prep course. And they are doing so without any guarantee — because no such guarantee exists, because law schools are not required to refund tuition to graduates who cannot pass the bar, because the ABA's response to below-threshold schools is to put them on notice, not to make their students whole.

The ABA's Accreditation Threat: A Toothless Warning

What happens to a law school that falls below the ABA's 75% ultimate bar pass rate threshold? In theory: consequences. The ABA's Standards and Rules of Procedure for Approval of Law Schools provide that a school falling below the threshold is not in compliance with Standard 316 and must come into compliance within a specified period or face accreditation consequences up to and including withdrawal of ABA approval.

In practice: very little. The ABA has taken steps against persistently failing schools over the years — it has issued show-cause orders, it has placed schools on probation, it has in rare cases withdrawn provisional accreditation from schools that couldn't get their bar pass rates up. But the process is slow, the schools' lawyers are skilled at buying time, and the ABA's appetite for actually revoking accreditation from established institutions is limited by the very relationships between the accrediting body and its member institutions that have always compromised the ABA's ability to police its own industry.

The result is a system in which schools that cannot reliably prepare their graduates to pass the bar — schools that, by any objective standard, are selling a defective product at unconscionable prices — retain full ABA accreditation, continue to recruit students, and continue to collect tuition. The warning that the ABA issues has the practical effect of a parking ticket: it registers the violation without changing the behavior.

Meanwhile, the students who enrolled at those schools while they were below the threshold — or while they were hovering just above it — have no recourse. There is no law school lemon law. There is no tuition refund for graduates who passed three years and $200,000 into an institution that the ABA has quietly flagged as non-compliant. There is no disclosure requirement that tells an incoming 1L: this school failed to meet the ABA's minimum bar passage standard last year, and you are about to pay full freight to attend it.

The Study That Proves Law School Pays — Using Data From Before the Internet Bubble Popped

On April 7, 2026, Reuters published a story about a new academic study from Yale economics professor Joseph Altonji and Vassar economics professor Zhengren Zhu. The headline: "Does a Law Degree Pay Off? New Study Says Yes." The finding: a law degree increases graduates' earnings by 59% on average — third-highest among advanced degrees, behind only medical and pharmacy degrees. When adjusted for the relative cost of law school, the earnings boost is 41%.

The legal education industry immediately circulated the study. It was the answer to the critics. The data. The proof. The rebuttal to everyone who had been saying that law school is a bad investment.

Here is what the Reuters headline did not mention: the study analyzes earnings data spanning from the early 1990s to 2018. The most recent data in the study predates ChatGPT by four years. It predates the mass-market rollout of AI legal tools by five years. It predates Anthropic CEO Dario Amodei's prediction — that AI will eliminate half of entry-level white-collar jobs within five years — by seven years. The data in the study that is supposed to prove law school pays off is from a world in which AI was a research topic, not an employment threat.

Let that sink in. The best the legal education industry can produce, in April 2026, to defend the economic case for a $150,000-to-$300,000 credential is a study whose data ends in 2018. In 2018, the iPhone X had just been released. In 2018, Amazon's Alexa was still mostly answering questions about the weather. In 2018, no major law firm had deployed an AI tool for substantive legal work. In 2018, the entry-level legal job market that the study's earnings data reflects was not yet competing with language models that can review contracts, draft motions, and synthesize case law in seconds.

The study is not asking whether a law degree pays off in 2026. It is asking whether a law degree paid off in a period that ended eight years ago. The answer to that historical question is yes. The answer to the question that actually matters — does paying $200,000 for a law degree in 2026 pay off in a legal job market that will be materially different by the time today's 1Ls graduate in 2029 — is not answered by the Altonji-Zhu study, because the Altonji-Zhu study does not contain data that could answer it.

The Admission Buried in the Study's Own Findings

The most damaging finding in the Altonji-Zhu study is not the headline number. It is the qualifier: "MBA and J.D. programs that ranked higher in U.S. News & World Report rankings yield higher returns than lower-ranking ones."

This is not a minor caveat. This is the thesis that the legal education industry cannot afford to advertise: the economic case for law school is strongest for the schools that admit the fewest students. The 99.43% bar pass rate is at Stanford, which accepts fewer than 10% of applicants. The 98.54% rate is at Yale, which accepts fewer than 8% of applicants. The "law degree pays off" headline is generated by the aggregate earnings of everyone who goes to law school — including the people who go to Stanford and Harvard and end up at BigLaw firms paying $225,000 a year and receive enormous returns on their educational investment.

Those earnings are averaging in with the earnings of graduates from schools ranked 100th, 150th, 180th — schools where the first-time bar pass rate might be 70% or 75% or 80%, where employment outcomes are substantially worse, where graduates enter regional legal markets where the median attorney salary is $70,000 to $90,000 a year. When you average those graduates' earnings with the Stanford and Harvard graduates' earnings, you get a number that sounds like "law school pays off." When you disaggregate the number by school tier, you get a picture that is considerably more complicated.

The Altonji-Zhu study says it itself: higher-ranked programs yield higher returns. What that means, for the prospective law student at the median school — the school ranked 100th, where tuition is $50,000 a year, where bar passage rates are in the 80s, where BigLaw hiring is minimal — is that the aggregate number is not her number. Her number is lower. The question is how much lower, and whether it justifies the investment, and the study's own data — ending in 2018, before AI transformed the calculus — cannot answer it.

What Happens to the Bar Failures

The ABA's 2025 bar passage data does not follow the graduates who failed the bar. It does not document what happens to the 8% of law graduates who cannot pass within two years, or the 42% of overall bar takers who fail on any given test. It does not track their debt. It does not record what they do for income when the legal career they borrowed $150,000 to pursue fails to materialize because they cannot clear the credential threshold that practice requires.

What we know comes from debt data and anecdote. The average law school graduate carries $137,000 in student loan debt. At 8.05% interest — the current federal Grad PLUS rate — a $137,000 balance accrues $11,029 in interest in the first year of repayment, whether or not the borrower has passed the bar, whether or not the borrower has found employment in law. The clock runs from the moment the loan is disbursed, not from the moment the degree pays off.

A graduate of a school below the ABA's 75% threshold who fails the bar on the first attempt has three options: study for and retake the bar while paying interest on six-figure debt and living on whatever income is available in the meantime; abandon the pursuit of legal practice and accept that the credential they borrowed $150,000 to obtain will not be the primary credential in their career; or find employment in a law-adjacent role that does not require bar passage and hope that the J.D. is a differentiating factor rather than an expensive credential that doesn't match the job description.

None of these options appear in the law school marketing materials. The glossy brochures show the students who passed. They show the attorneys who thrived. They do not show the 8% who couldn't clear the bar within two years, or the repeat takers who bring the overall pass rate to 58%, or the graduates of the six below-threshold schools who are paying interest on loans for a credential that has not yet delivered what it promised.

The Bimodal Salary Distribution Nobody Explains to Applicants

The legal industry's salary data has a structural feature that law school marketing materials systematically obscure: it is bimodal. Attorney salaries in the United States do not form a bell curve with a comfortable median. They cluster at two poles — and the poles are far apart.

The first cluster is BigLaw: the large national and international firms whose starting associate salaries are $225,000. The Cravath scale, as it is known in the legal industry, sets the compensation floor for major firm associates. Getting into a BigLaw firm requires graduating from a highly ranked law school, typically in the top of your class, with credentials that position you for offers at firms that interview almost exclusively at T14 institutions or near the top of regional schools.

The second cluster is everything else: government, small firms, solo practice, public interest, in-house work at companies that aren't paying law firm rates. Salaries in this cluster typically range from $55,000 to $95,000 depending on geography and employer type. The National Association for Law Placement's employment data has documented this bimodal distribution for years.

When the Altonji-Zhu study calculates that a law degree increases earnings by 59% on average, it is averaging together the BigLaw associates making $225,000 and the public defenders making $58,000. The average looks reassuring. The distribution it conceals is not.

For the student at the median law school — a school ranked in the 80s or 100s, charging $50,000 a year in tuition, with a bar passage rate somewhere in the mid-80s — the realistic salary expectation is not $225,000. It is not the headline number that drives the Altonji-Zhu study's "59% earnings increase" finding. The realistic salary is somewhere in the $70,000 to $90,000 range, in a market where AI is beginning to compress demand for the entry-level work that those salaries are supposed to reflect.

A $137,000 debt load at 8.05% interest, repaid over ten years, requires approximately $1,670 per month in loan payments. After tax, an attorney making $80,000 a year — the median outside BigLaw in many markets — takes home approximately $4,800 per month. After student loan payments, $3,130 per month remains for housing, food, transportation, and everything else. In New York, San Francisco, Los Angeles, Chicago, or any other city where legal employers cluster, $3,130 per month after loan payments is poverty wages for a professional who spent seven years in post-secondary education and borrowed $137,000 to get there.

This math is not hidden. It is just not in the law school brochure.

The Historical Data Problem and the AI Cliff

The most important number that the Altonji-Zhu study cannot provide is what happens to the law degree's earning premium after 2026. The study's data ends in 2018. Its findings reflect a legal job market that has been under AI disruption for, at most, two years — a period so early that the structural effects have barely registered in employment statistics.

The disruption timeline matters. The Dallas Federal Reserve Bank published a working paper in early 2026 documenting that AI adoption in legal work is already reducing entry-level hiring in AI-exposed occupations while raising wages for experienced workers who manage AI tools. The effect is a profession that is bifurcating: fewer entry-level jobs, higher compensation for senior attorneys who use AI to do more with less human labor. The entry ramp — the first years of associate work, the training period, the billable hours that justify entry-level salaries — is compressing.

Anthropic CEO Dario Amodei predicted, in May 2025, that AI will eliminate half of entry-level white-collar jobs — including legal work — within five years. His window runs to 2030. Law students enrolling this fall will graduate in 2029. They are borrowing $150,000 to $300,000 against a job market that, if Amodei's projection is anywhere near correct, will have half as many entry-level positions available as the job market reflected in the Altonji-Zhu study's 2018 data.

The "law school pays off" study cannot account for this because it does not contain data from this period. It is measuring the return on a degree in a profession as it existed before the most significant technological disruption in the profession's history. Using it to justify enrolling in law school in 2026 is like using 2005 travel agent salary data to argue that becoming a travel agent is a good career choice.

The Accreditation System That Protects Schools, Not Students

The ABA's accreditation system is supposed to be the consumer protection mechanism for law school students. The idea is that ABA approval signals quality: a school with the ABA's stamp meets the standards the profession has set for legal education, and graduates of ABA-approved schools can sit for the bar in any state in the country.

What the accreditation system actually does is protect the institutions. The ABA's accreditation standards cover facilities, library resources, faculty qualifications, clinical programs, and — as we have seen — bar passage rates. They do not cover prospective risk disclosure. They do not require schools to tell applicants: here is what the job market will look like when you graduate in three years, given current AI adoption trends. They do not require schools to disclose: our bar passage rate last year was below the ABA's minimum accreditation standard, and you should factor that into your decision about paying $200,000 to attend.

The Standard 509 disclosure requirements — the ABA's mechanism for required transparency — cover historical outcomes. What did last year's graduates find? What did the bar passage rate look like last year? These are the questions the ABA requires schools to answer. The question that actually matters — what will the legal job market look like when this year's incoming class graduates in 2029, and is the credential worth the price — is not a question the ABA requires anyone to answer.

The result is an accreditation system that the ABA describes as consumer protection and that functions, in practice, as an institutional certification that enables schools to charge premium prices without premium accountability. A school with ABA accreditation can continue recruiting students and collecting tuition even while failing to meet the bar passage standard that the ABA has established as the floor of acceptable outcomes. The consequences — for institutions — are slow, negotiated, and rarely terminal. The consequences — for students — are immediate, financial, and permanent.

The Two Numbers That Tell the Whole Story

Here are the two numbers that the legal education industry's self-congratulatory bar data release and its "law school pays off" study cannot be made to add up:

Number one: 99.43%. That is Stanford Law School's first-time bar passage rate in 2025. It is remarkable. It reflects a school that admits students with LSAT scores in the 99th percentile, that graduates students into BigLaw and federal clerkships, that produces alumni who define American legal practice at its highest levels. A Stanford J.D. is worth what Stanford charges for it, for the students who get in and who achieve what Stanford's alumni achieve.

Number two: 75%. That is the ABA's minimum bar passage standard — and six schools fell below it. Those six schools are also ABA-accredited. They also have the ABA's stamp. Their graduates can also sit for the bar in any state in the country. Their degrees also cost $150,000 to $250,000. There is no asterisk on those schools' accreditation that tells a prospective student: this institution failed to meet the minimum standard the ABA has set for acceptable educational outcomes.

The gap between 99.43% and 75% is the gap between the credential that law school marketing sells and the credential that a significant portion of law school graduates receive. The industry averages these schools together and produces a national pass rate that sounds acceptable — 84% first-time, 92% over two years. The aggregate conceals what the distribution reveals: a profession with a small number of elite institutions producing excellent outcomes and a much larger number of institutions charging comparable prices for substantially inferior results.

The Altonji-Zhu study tells us that higher-ranked programs yield higher returns than lower-ranked ones. The ABA data tells us that higher-ranked programs also produce more bar-passing graduates than lower-ranked ones. Neither of these findings is surprising. What is surprising — or should be — is that both findings are treated as background noise by an industry that markets a homogeneous product at homogeneous prices to students who have not been given the information they would need to distinguish the product they are actually buying from the product they are being sold.

The Honest Disclosure the ABA Should Require But Won't

An honest law school disclosure document, in 2026, would say something like the following:

This institution is ABA-accredited. Our first-time bar passage rate in 2025 was [X]%. [If below 84%: This is below the national average. If below 75%: This is below the ABA's minimum accreditation standard.] Our graduates who take the bar pass at a rate of [Y]% within two years. The median starting salary for our graduates who find employment requiring bar passage is [Z] per year. Approximately [A]% of our graduates find employment in positions that require a J.D. within ten months of graduation. The median debt for our graduates is [$B]. We cannot guarantee bar passage. We cannot guarantee employment. We cannot predict what the legal job market will look like in three years, when you graduate. We are currently collecting tuition at a time when the CEO of one of the world's leading AI companies has predicted that artificial intelligence will eliminate half of entry-level legal jobs within five years. You should factor this information into your decision.

No law school publishes this document. No law school is required to. The ABA's Standard 509 disclosures require schools to publish their historical outcomes in formats that legal education researchers can decode — but that 22-year-old applicants, without training in how to read and interpret the data, are unlikely to process correctly before the application deadline passes and the enrollment deposit is due.

The industry produces the aggregate and calls it transparency. The aggregate is accurate and the transparency is real in a technical sense. What is missing is context, explanation, and the kind of plain-language disclosure that would allow a first-generation college graduate — who has no family attorney, no law school alumni in the family, no network of people who have navigated these decisions before — to make a genuinely informed choice.

What the Numbers Say About 2029

The students applying to law school right now will graduate in 2029. The bar passage data from 2025 is, for them, historical. The job market data from 2018 — the most recent data in the Altonji-Zhu study — is ancient history. What matters for them is 2029: what will the bar look like then, what will the legal job market look like then, and what will a law degree from their school be worth in that environment.

The ABA's 2025 bar data cannot answer that question. The Altonji-Zhu study cannot answer that question. The schools' marketing materials, which are built on historical data and optimistic projections, cannot answer that question honestly.

What we can say, with confidence, is this: six ABA-accredited schools could not get three-quarters of their 2023 graduates past the bar within two years. The national first-time pass rate is 84%, which means one in six first-time takers fails. The overall pass rate — including repeaters — is 58%. Average graduate debt is $137,000. The best study the legal education industry can produce to argue that law school pays off uses data ending in 2018. AI is already reducing entry-level legal hiring in documented studies. The CEO of Anthropic says half of entry-level legal jobs will be gone within five years.

The students enrolling in law school this fall are betting $150,000 to $300,000 that these numbers don't apply to them. They are betting that they will beat the bar pass rate, that they will find the job, that the AI disruption will not reach their niche, that the credential will be worth the price in the market they will enter in 2029.

Some of them will be right. The ones who go to Stanford and Yale and Harvard, who score in the 99th percentile on the LSAT, who graduate in the top of their classes and land BigLaw offers — they will almost certainly be right. Their debt will be manageable relative to their income. Their bar passage rate will be near 100%. Their AI risk is lower because the most senior, complex, judgment-intensive legal work is the work that AI is slowest to displace.

The rest — the ones going to the school ranked 100th, the ones at institutions below the ABA's 75% threshold, the ones without the LSAT scores for T14 admission, the ones who are told by their pre-law advisor that any ABA-accredited school is a good investment — are rolling a much harder set of dice. And the industry is handing them a set of marketing materials built on aggregate statistics, an eight-year-old study, and a bar passage headline that averages Stanford's 99.43% with six schools that can't clear 75%.

That is not consumer protection. It is a sales pitch dressed up as data.

---

Sources and Citations

• Reuters / Sloan, K. (Mar. 12, 2026). "These US law schools aced the bar exam in 2025." reuters.com
• TaxProf Blog. (Mar. 15, 2026). "New ABA Bar Data: 1-Year Pass Rate Increased 1 Percentage Point; 7 Law Schools Failed 2-Year 75% Pass Rate Accreditation Threshold." taxprofblog.aals.org
• Reuters / Sloan, K. (Apr. 7, 2026). "Does a law degree pay off? New study says yes." reuters.com
• Altonji, J. & Zhu, Z. (2026). "Do Graduate Degrees Pay Off?" Postsecondary Education & Economics Research Center. peer-center.org
• ELFI. (2026). "How Much Does Going to Law School Cost?" elfi.com
• Research.com. (2026). "How Many Years of College Does It Take to Become a Lawyer?" research.com
• Missouri Lawyers Media. (Apr. 10, 2026). "Missouri bar exam pass rate declines slightly in latest results." molawyersmedia.com
• ABA Journal. (Apr. 8, 2026). "Enrollment in law school of first-gen college grads drops, new LSAC report finds." abajournal.com
• Dallas Federal Reserve Bank. (Feb. 2026). Working paper on AI and entry-level hiring in AI-exposed occupations.
• Anthropic / Amodei, D. (May 2025). Public remarks on AI and white-collar employment disruption.
• ABA Standards and Rules of Procedure for Approval of Law Schools, Standard 316 (Bar Passage).