Eric Hanushek on Test-based Accountability, Federal Funding, and School Finance 08/10/2009
What this covers
Eric Hanushek of Stanford's Hoover Institution and Russ Roberts examine how to measure and improve American education through the lens of school finance and accountability policy. Hanushek presents empirical findings from his research into whether court-ordered spending increases actually raised student achievement, while Roberts pushes back with questions about testing regimes and the role of government in schooling. The conversation moves across the impact of No Child Left Behind, the mixed results of the charter school movement, and the mechanics of how school finance should connect to performance rather than simply inputs.
The core argument turns on a claim about causation: that billions pumped into schools through major court cases in New Jersey, Kentucky, and Wyoming produced little measurable gain in student outcomes relative to national trends. Hanushek argues that money divorced from incentives and accountability has failed, and proposes instead linking finance directly to measured learning outcomes, rewarding performance, and devolving control closer to parents and localities. Along the way, the conversation addresses why test-based accountability persists despite its unpopular branding, whether charter schools are genuinely serving disadvantaged students or merely skimming high performers, and how technology might eventually disrupt traditional schooling's rigidities. Roberts introduces a more radical position: that government should step back entirely from providing education, or at minimum let dollars follow students rather than schools. The discussion also examines whether courts have any legitimate role setting education budgets, how union contracts constrain what schools can do, and whether adaptive testing could better measure the full range of student achievement rather than only proficiency floors.
Hanushek argues that school funding decided by courts and measured in dollar inputs has failed to improve student outcomes, and that education systems should instead tie funding and policy together around measured learning outcomes, incentives, and local control.
- Major adequacy court cases (NJ, KY, WY) pumped billions in with little measurable achievement gain relative to other states
- Money as an input is a poor index of school quality; incentives and accountability matter more
- Effective reform requires linking finance to policy, rewarding performance, and restoring local/parental control
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.
Evaluating NCLB's impact is extraordinarily difficult because by 2002, 47 states already had some form of test-based accountability in practice, leaving no clean control group to compare the federal treatment against.
“when it went into effect in 2002 47 states already had some form of test based accountability and practice so it's hard to know what's the control group”
In the major adequacy court cases—New Jersey, Kentucky, and Wyoming—that pumped billions of dollars into schools, there is very little evidence that students did any better relative to other states without such cases; the only place where performance improved (Massachusetts) also implemented a series of structural reforms, suggesting more money in a better system has effect but money alone does not.
“in the large court cases and court rulings that pumped billions of dollars into New Jersey Kentucky and Wyoming there's very little evidence that the kids did any better relative to other states”
Adaptive testing—like the GRE, where a screening test gauges a student's level and then asks questions within their relevant range—could allow accountability tests to measure the full distribution of achievement rather than only bottom-end proficiency, but little effort has gone into developing this for school accountability.
“you can have adaptive testing so that you give kids a screening test it's limited number of questions that gives some idea of what level of knowledge they have and then you ask us sir questions within the range that is relevant to them”
Starting around 1970, advocates used state constitutions requiring states to fund schools to bring 'equity' lawsuits (beginning with Serrano v. Priest in California) arguing that locally-raised funding produced inequitable spending variation across districts; after about 20 years these wound down and advocates shifted to 'adequacy' suits arguing constitutions required more funding to make schools adequate.
“starting in about 1970 there were people who wanted to get the courts involved in the distribution of funding across local districts the argument was simply inequity”
New technology could revolutionize the politics of schooling because it is hard to unionize a dispersed technological engine delivering education, potentially changing what is taught and how—but whether this happens depends on whether the politics block or permit technology's introduction.
“new technology could revolutionize the politics of schools because it's hard to unionize a dispersed technological engine that that's providing education”
In roughly 35–40 years of school-finance court cases, essentially no one had previously bothered to look at whether the court-ordered money actually improved outcomes—Hanushek's book did this novel check using NAEP growth in reading and math at fourth and eighth grade relative to national growth.
“in 35 to 40 years of these court cases there is a couple of little leaven nobody bothered the looker door nobody wanted the look”
A major national CREDO study found charter schools split into three groups: a set doing amazingly well, a set indistinguishable from the traditional schools that feed them, and a third—larger than the high-performing group—that is doing badly, often because the founders lacked administrative or educational skills.
“there are a group of charter schools that are doing amazingly well there is a group that's indistinguishable on average from the traditional schools that feed them and then there's another group that's actually larger than the group that's doing well it's doing badly”
A well-educated, high-achieving workforce makes a large difference to the functioning of the US economy, so future economic well-being is partly dependent on whether children actually learn in the classroom—and US students currently lag those in Finland and other places.
“a well-educated at i achieving workforce means a big it makes a big difference in terms of the functioning of the US economy”
No Child Left Behind represented a deeper federal intrusion into education policy itself—specifying what kids should know and what states must do to receive compensatory funds—rather than the federal government's traditional role of merely funding specialized programs for disadvantaged and special-education students.
“with President George Bush's first major domestic legislation the federal government got deeper involved into policy itself by specifying what kids should know and what the states had to do in order to get their compensatory money”
Many distinctive practices of successful charter chains like KIPP—longer school days and years, Saturday programs, teachers taking calls at all hours, enforcing behavior norms—are things public schools cannot easily do not because of pedagogy but because they violate union contracts that specify contact hours and out-of-classroom duties in great detail.
“some of these things are violations of current union contracts like taking phone calls at night... working more than three hours in the day”
Because public schooling is given away with no out-of-pocket cost to parents, parents pay nothing directly, which substantially reduces their involvement; full state funding (as in California) further removes any local ability to reward or punish schools through funding, contributing to California being the 47th-ranked state on NAEP performance.
“the fact that it's given away with no out of pocket to the public students so the parents pay nothing out of pocket and I think that has a big impact on their involvement”
Classroom schooling is strikingly inefficient—a large proportion of time is spent keeping children well-behaved—such that a knowledgeable parent could plausibly teach in two to four hours what a child learns in a full school day, suggesting much of school's value may be socialization rather than education.
“what strikes me is how inefficient classroom schooling is the proportion of time that's spent keeping kids well-behaved”
A better system requires not separating finance from policy and ensuring that finance rewards good performance rather than bad: an accountability system measuring learning, a base of sufficient and equitable resources (with more for harder-to-educate students), additional funding directed to schools, teachers, and administrators who perform better, and crucially more local control and parental involvement through choice.
“we shouldn't try to separate finance from policy... your policies and your finance have to reward good performance and not bad performance”
Performance pay for teachers should not be set at the state capital because salary schedules set there are out of touch with districts; instead the state could provide a pot of money to support locally-negotiated performance pay that meets parameters such as a substantial portion being based on objective learning data.
“you don't want to set salary schedules in the state capital because they're just out of touch with what's going on in districts... you can envision the system where the California State Legislature had a pot of money that could go to districts to support the bonus part performance part of a salary schedule if it was negotiated locally”
New Jersey's Abbott districts (31 of 500+) were given essentially unlimited spending capacity—merely needing to ask the court for more money for a given purpose—with court cases dictating preschool provision and class sizes; qualification required both large concentrations of poor kids and historically low achievement, perversely rewarding poor performance with unlimited funds.
“there were 31 districts in New Jersey that are called Abbott districts... they were given essentially unlimited spending in capacity all they had to do is come into the courts and say we need more money for X and the court said oh I we we agree”
Wyoming, around the middle of states in per-pupil spending before its 1996 court case, shot up to the top five spending states afterward—illustrating how courts ordering states to be 'visionary and unsurpassed' creates a perverse race among states to spend ever more.
“Wyoming is a top five spending state in the nation... it happened after this court kid which was in 1996... they were basically told in a court case be visionary and unsurpassed”
Contrary to early fears that charter schools would 'cream off' good students, they have tended to serve a very disadvantaged, central-city population—precisely those who traditionally lacked the school choice that suburban families exercise by moving districts.
“charter schools have tended to serve a very disadvantaged population they tend to be much more in central cities and out in the suburbs creaming off the good kids which is what get one of the worries was when it was started”
Disaggregating by race shows the same disappointing story: black students in the court-case states did not do as well as the nation in general, even though disadvantaged children were always identified as the intended recipients of the additional court-ordered funds.
“if you disaggregate by race of the students you find the same story in particular black kids in these states didn't do as well as the nation in general even though they were always identified as the recipients”
It is almost inconceivable that the US will do away with test-based accountability in some form, because parents like it and it has been a productive policy, even though the 'No Child Left Behind' brand name is among the worst in the country and will not survive.
“it's almost inconceivable to me that we will do away with test based accountability in some form parents like it it's been I think a productive policy”
Roberts's preferred solution would be to get government out of schooling entirely or at least out of provision—keeping perhaps a subsidy role with dollars following students not schools—but with 90% of the population in public schools, a free and open voucher system is too far off, so the realistic path is more choice via expanded charters and market-like incentives supported by government.
“my preference would be to get government of schooling entirely or at least certainly have the provision of schooling maybe you could have some sort of subsidy role... where dollars would follow students not schools”
NCLB largely codified at the federal level what a vast majority of states had already adopted on their own—test-based accountability—so it was a continuation that put a federal stamp on existing state practice rather than a wholly new policy.
“that law actually just put in place something that had been done by a vast majority of the states before the law went into effect so it's a continuation and it's putting a federal stamp on that”
Some learning tasks—such as drilling fourth-grade math problems—are probably done better by a computer than by even most good math teachers, and certainly more cheaply in both money and time, so the argument for using technology for such tasks is overwhelming.
“drilling on math problems in the fourth grade is probably better done by a computer than it is by the math teacher and even by even most of the good math teachers”
Courts deciding how much should be appropriated for schools—a function traditionally reserved to legislatures—is shocking in a democracy; courts are not well suited to make these policies and lack the capacity, even setting aside constitutional interpretation.
“deciding how much should be appropriated for schools which is shocking really in a democracy that that the courts took that”
Compensatory funding from states to districts with low fiscal capacity (practiced since the 1920s) makes sense to ameliorate tax-base discrepancies so people have a more equitable ability to raise funds, though full neutralization of district differences may be undesirable for incentive purposes.
“since the 1920s almost every state in the Union has in fact tried to make up for districts that don't have much fiscal capacity... to ameliorate some of the discrepancies we have in the tax bases across districts”
Carrot-and-stick funding that punishes bad policy and rewards good policy is hard to do—the World Bank's similar approach of withholding money from countries with bad policies has not worked very well—though it might work better in the domestic context.
“the World Bank hasn't been very successful at it right they have a similar idea that we'll will punish people who have bad policies and don't give money... it doesn't seem to work very well”
The 2009 federal Supreme Court case Flores v. Arizona (English language learners) produced a 5-4 majority strongly stating that courts should pay attention to student performance rather than how much money is spent, citing Hanushek's work and new book—offering optimism that adequacy litigation may face more skeptical judicial reasoning.
“it went to the Supreme Court in this spring spring of 2009 and the five to four majority of the court came out with a very strongly worded statement that said we had to pay attention to student performance and not how much money we put in”
An open question for economists is whether parental choice alone is sufficient to squeeze out the failing bottom tier of charter schools, or whether a regulatory system that re-evaluates charters and revokes them is needed.
“the open question for economists is whether parental choice is sufficient to in fact squeeze out this bottom end or whether we want to go back to a regulatory system that somehow re-evaluates their charter”
States with strong accountability systems in the 1990s showed better achievement on reading and math tests, and tracing individual states before and after 2002 shows achievement increases were somewhat larger after 2002—offering limited evidence that accountability improves performance.
“states that had strong accountability systems did better achievement test reading and math tests of kids there's also a little bit of evidence that if you trace individual state performance before and after 2002 that the increases in achievement were larger after 2002”
Public interest in education quality has remained remarkably high and consistent since at least the 1983 'A Nation at Risk' report, and education 'sells' politically—evidenced by the president labeling $100 billion of stimulus as education even though much of it is bricks and mortar with little to do with learning.
“a consistent public interest in schools that has traced back at least in 1983 when there was a major government report called a nation at risk”
Variation in charter school presence across cities is driven mainly by state regulations—nine or ten states allow no charters, and many that do impose caps on total numbers or restrictions on charter chains—and secondarily by entrepreneurial activity.
“there are nine or ten states that don't allow any charter schools at the current time and of those that allow charter schools many of them have caps on the total number”
Although states could legally refuse to comply with NCLB, giving up roughly 10% of funding—amounting to 50–60 billion dollars nationally—is more than any state is willing to forgo, which is why threats to opt out have not materialized.
“giving up 10% of funding is more than any state wants to do because these amount to a big number I mean we're talking about 50 or 60 billion dollars total for the nation”
Federal dollars constitute about nine percent of total education expenditures today, up from six or seven percent ten years ago, with states funding about half, local governments about forty percent, and the federal government about ten percent—so federal funding, while a large absolute number (50–60 billion), is a relatively small share.
“right now about nine percent of total expenditures are from federal dollars that's up from six or seven percent ten years ago”
The federal stimulus 'Race to the Top' fund ($4.35 billion at the Secretary of Education's discretion) would withhold money from states that cannot track student performance or link it to schools and teachers, or that restrict charter schools and choice, and reward states experimenting with teacher performance pay—an experiment whose effects will be observable in coming years.
“the federal stimulus money to education has a portion of it four point three five billion dollars that gives discretion to the Secretary of Education... I will not give this the money... to states that can't track the performance of students... that restrict charter schools”
The two testing companies that produce all the accountability tests write 50 different versions and sell them as new to 50 states, yet the tests have great similarities because fourth-grade math in Florida looks much like fourth-grade math in New York, suggesting state-specific test design is largely wasteful.
“insisting that the two testing companies that make all the tests write 50 different versions and sell them as if they're new things to 50 states”
Charter schools provide, in general, more safety for kids, which is something parents value highly even if such schools score poorly on basic proficiency tests.
“one of the things that we know that they're doing is providing in general more safety for kids”
Charter schools experience large turnover of students both coming and going, which is one of the difficulties they face beyond startup, and parents can freely move a child out of a poorly-performing charter back into the neighborhood public school.
“charter schools have a large turnover of kids both coming in and out which is one of the difficulties that they face not only just starting up but facing a population that's turning over a lot”
Suburban schools where everyone passes the proficiency tests can eliminate the distortions potentially caused by NCLB by adopting their own higher standards rather than relying solely on the federal proficiency floor.
“the suburban schools don't have to rely entirely on that if they know that that's everybody can pass that they ought to have other standards and they can in fact eliminate the distortions that are potentially caused by NCLB by putting in their own policies”
Charter schools and homeschooling each serve roughly the same small share of students—about two to three percent—compared to roughly ten percent in private schools overall, so neither is a large proportion of the total.
“two or three percent of the population in charters and two or three percent and homeschooling... total private schools are ten percent”