What this covers
Eric Hanushek, a Senior Fellow at the Hoover Institution, presents research on why massive increases in school spending have failed to improve student achievement in the United States and discusses the implications for economic growth at home and abroad. Russ Roberts, host of EconTalk, guides the conversation through Hanushek's findings on the relationship between educational quality and economic outcomes. The core argument rests on a stark empirical claim: real per-pupil spending has more than tripled since 1960, yet test scores in reading, math, and science have remained essentially flat from 1970 onward, despite implementation of the exact reforms experts advocated—smaller classes, more experienced teachers, more advanced degrees among staff.
Hanushek argues that spending has flowed into measurable but ineffective inputs because schools have directed resources toward factors that sound reasonable but don't predict teaching quality: class size, teacher credentials, and years of experience. The Tennessee STAR experiment found only modest, temporary gains from reducing class sizes, and systematic studies using value-added methods show that teacher quality varies enormously yet is uncorrelated with master's degrees or experience beyond the first couple of years. The discussion spans why individual teacher quality matters far more than these credentials, why public schools struggle to identify and retain effective teachers, and how institutional structures—tenure, single salary schedules, union contracts, and the absence of school choice—insulate administrators from the information and incentives needed to manage talent effectively. The conversation also addresses charter schools, accountability testing, and merit pay, touching on why competition and market mechanisms might work where current institutional arrangements have not.
Hanushek argues that decades of dramatically increased school spending have produced flat student achievement because resources have flowed into measurable inputs (class size, teacher credentials, experience) that don't predict teaching quality, while the real driver—individual teacher quality—goes unrewarded due to union contracts, tenure, and the absence of market incentives.
- Real per-pupil spending more than tripled since 1960 while test scores stayed flat from 1970 onward
- Teacher quality varies enormously but is uncorrelated with master's degrees or experience past the first year or two
- Public school institutional structures (tenure, single salary schedule, no competition) insulate schools from the information needed to retain good teachers and remove bad ones
This asset isn't compiled yet
You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.
A low-income student who has a good teacher (around the 85th percentile rather than the 50th) three or four years in a row can be expected to overcome the average achievement gap associated with coming from a low-income family, meaning teacher quality can compensate for family-background disadvantage.
“if we look at a student from a low-income family and imagine that this student had one of these good teachers three years in a row or maybe four years in a row compared to an average teacher the gains with that good teacher would be expected to overcome the average difference”
Public school principals are not residual claimants and face no competition, so they keep their jobs regardless of school performance and lack the incentive to care about outcome quality.
“they're not the residual claimant the phrase we use in economics for the owner of a business who captures some of the gains when it's run well... The principal is going to keep his or her job regardless of the performance of the school”
Despite tripled spending, US student test scores in reading, math, and science have been essentially flat from 1970 to today, even though the exact reforms experts called for—smaller class sizes, more experienced teachers, more teachers with master's degrees—were implemented.
“test scores of students have been virtually constant or flat from 1970 through today”
Whether someone is a good teacher is virtually uncorrelated with whether they hold a master's degree or have more experience, except for the first year or two when a brief break-in period matters; after that there is no general increase in teaching performance with experience.
“who is a good teacher is virtually uncorrelated with whether they have graduate education and a master's degree or whether they have more experience save perhaps in the first year”
The Tennessee STAR random-assignment class-size experiment of the mid-1980s—comparing classes of about 23-24 students to 15—found at best small, modest gains in kindergarten or first grade that did not compound in subsequent years, undermining class-size reduction as a reliable lever.
“the state of Tennessee before they put more money into reducing class sizes decided to run an actual experiment where they randomly assigned some students to small classes in some of the larger classes”
Studies following thousands of teachers over time using value-added methods (measuring how much achievement growth students gain in a given year) show that some teachers consistently produce high growth and others consistently low growth, and these are substantial, reliable differences—yet uncorrelated with education level or experience.
“when we follow teachers over time is that some teachers year after year get high growth and student achievement it's not the high level it's given where they start how far do they progress”
Principals, teachers, parents, and even janitors can reliably identify teachers at the extremes of the quality distribution (the best and worst), even though there is confusion in the middle—yet the public school system is structured to insulate school operations from acting on that information.
“almost everybody who walks into a school can tell you who is at the extremes in that there's very little difference of opinion on this the principals can tell you that the teachers can tell you that the parents can tell you that the janitors can tell you”
If a test were well-designed to cover the full range of material we care about, then 'teaching to the test' is not a problem—it is exactly what we want teachers to do, just as college professors teach to their exams; the problem is that many current US accountability tests are too simplistic and low-level.
“if we had a good test which ranged that covered the range of material that we were interested in then there's nothing wrong with teaching to the test actually exactly what we want the teachers to do”
Because we cannot identify good teachers from observable credentials, the operationally useful assumption is that teaching skill is largely innate, so policy should focus on keeping good teachers in the classroom and removing bad ones rather than trying to retrain teachers through new education-school courses.
“my own current view for just for operational purposes is to assume that it's something innate in individuals that some people are born to be good at it”
Private schools tend not to use widely varying salaries but are much more careful about whom they hire and retain, whereas in public schools most retention decisions are effectively made by teachers themselves (whether they choose to stay), which makes a large difference in quality outcomes.
“private schools don't have really widely varying salaries... but they are much more careful about who they keep and who they retain so that they make their decisions often on hiring and retention”
Cheating on accountability tests is largely self-correcting because a teacher who inflates scores one year must cheat as much or more the next year to keep showing gains, so with basic safeguards cheating is not a major problem.
“you can cheat one year but then the next year when you're evaluated you have to cheat as much or more the next year to look better and so in my own view is that there with basic systems in place that cheating is not such a big issue”
A merit-pay system would be a disaster if principals themselves were not also subject to performance rewards, because teachers fear that evaluations would otherwise be based on politics and personal relationships rather than student performance; thus principals must have incentive systems tied to overall school performance and the freedom to make decisions.
“it would be a disaster to run a merit pay system with big rewards for performance in which the principals themselves were not subject to also performance rewards because that's just what teachers fear”
Collective bargaining has produced a 'single salary schedule' that pays teachers solely on a matrix of experience and graduate education, ignoring specialty (e.g., scarce math teachers) and quality, and has also led unions to bargain over classroom time, class size, and senior teachers' first choice of placement regardless of principal preference.
“the single salary schedule that is it's a look up the matrix that has years of experience and amount of graduate education... math teachers don't appear in this matrix all teachers get paid the same for experience”
The core problem is that public schools are insulated from competition: requiring students to attend the neighborhood school removes the market forces and information-processing institutions that normally allow consumers to exercise oversight, so most reform proposals are workarounds for that missing market discipline.
“the idea that you have to go to the school in your neighborhood if you want the free schooling... is a bizarre concept and a lot of these solutions we're trying to come up with our ways of trying to given that there's no inherent market forces from the consumer”
Real per-pupil educational spending in America has more than tripled since 1960 after adjusting for inflation, contradicting the common belief that schools have been starved of funds.
“since 1960 the real spending after we adjusted for inflation per pupil has more than tripled tripled in real terms”
Beyond the Tennessee experiment, a wide range of simple to sophisticated analyses controlling for family background and teacher differences shows no systematic relationship between class size and student performance.
“a wide range of simple to very sophisticated analyses of class size that essentially showed that there's no systematic relationship between class size and student performance controlling for other factors”
Hanushek's ideal merit system would not be based purely on mechanical test scores but would combine subjective information from principals and instructional leaders with some test data, because teachers contribute a broad range of activities to a school and test scores are needed primarily to enable comparison across schools where absolute quality is otherwise hard to judge.
“my ideal system would not have it based entirely on test scores and a mechanical system but in fact I would have subjective information from the principal or other instructional leaders”
Short-run feasible improvements include the accountability movement (valuable for shining a spotlight on problems even where institutions resist change) and the expansion of school choice through charter schools, though charters are individual 'mom-and-pop' institutions without consistent structure that work in some places and not others.
“I'm a big supporter of the current accountability movement in the u.s. because shining a spotlight on problems I think is very important”
Education differs from competitive markets like the market for cars because it is largely run by government and not subject to market forces, so additional spending often improves the happiness of people within the system without translating into better student performance.
“in an education it's largely run by the government and is not subject to many market forces so we see that things happen that improve the happiness of the people in education but that don't necessarily translate into better performance of students”
Teacher tenure (granted as early as the second year in states like California) and rigid contracts that reward only experience and graduate education—the very factors found not to matter for teaching—protect ineffective teachers from management action.
“We have tenure for teachers in some states such as California as early as the second year of teaching somebody has tenure so that protects people that aren't very good”
Competition not only sorts good schools from bad but also checks the power of teachers' unions, because when clients can go elsewhere unions cannot dictate terms without consequence; absent competition, union power over contracts and rules is unchecked.
“one of the things that competition does at least if it's outside competition where clients can go elsewhere is to limit the power the teachers unions can have... their power is checked if they don't have any students showing up at the door”
First-year teachers require a little training or break-in period, after which experience yields no further systematic performance gains.
“first year teachers require a little training or a little break in period but after the first year you don't see much that happens”
Teachers' unions fight 'tooth and nail' to stop or inhibit any expansion of school choice, including charter schools.
“they certainly fight tooth and nail for to stop any choice including charter schools and to try to inhibit any choice there”
Increases in special education spending, while worthwhile because they extend education to students previously excluded, cannot explain the tripling of per-pupil spending.
“the amount of increase in special education simply can't explain a tripling of spending”