Most voters have no sense of how much we actually spend on K–12 education. In 2007, Education Next and Harvard’s Program on Education Policy and Governance conducted a survey on the matter. Ninety percent of respondents underestimated per-pupil expenditures in their district, and most did so severely: The median estimate was $2,000 per pupil; in the districts surveyed, actual per-pupil spending varied in 2004–05 between $5,644 and $24,939, and the average was $10,377. On teacher salaries, respondents were closer to the mark: They underestimated salaries by only $14,370 on average.

As Americans grimly contemplate a decade-long economic slump, the time has come to take a long, hard look at these expenditures, which are not only high but rising. We should start by asking: Given that today it takes roughly as much time to teach a student how to read or how to solve an algebra equation as it did in 1980, why does it cost far more? In part because teachers’ unions have been so successful at lobbying for higher salaries. But their lobbying against measures that increase productivity may be an even bigger factor.

At the heart of the problem is Baumol’s cost disease, so named because it was identified by New York University economist William Baumol. Baumol noted that some sectors of the economy become more productive over time because of technological advances, while others stagnate — and that, counterintuitively, wages and salaries tend to rise in both types of sectors. He observed that when sectors experience productivity growth and raise their pay, they force the rest of the economy to follow suit to compete for workers. Babysitters and caregivers in nursing homes are textbook examples: Their work doesn’t get much less labor-intensive as technology marches forward, but if they are not to move to higher-paying sectors of the economy, their wages must keep pace with wages elsewhere.

The only solution to this problem is to increase productivity in the stagnant sectors. As economists Jack E. Triplett and Barry Bosworth wrote in 2002, this happened in the late 1990s in the service sector. As retail firms embraced the use of information technology, they found they could do more with less. Walmart was an essential driver of this retail revolution, replacing small, inefficient mom-and-pops with a sprawling and sophisticated logistical infrastructure that rivals that of the U.S. military. Inevitably, Walmart forced all of its competitors and all of its suppliers to improve their productivity, lest they be wiped out. Triplett and Bosworth even declared that Baumol’s cost disease had been cured. Unfortunately, the productivity boom in retail has slowed down, as all of the low-hanging fruit has been picked. Further productivity increases will likely prove more expensive, and more difficult, to achieve.

But because educational productivity has been so stagnant for so long, increasing productivity in education is well within our reach, provided we have the political will. The first step is simply explaining to cost-conscious voters why education is so expensive. But the next steps are harder, for they involve taking on the education cartel:

Choice. Earlier this year, the University of Arkansas Department of Education Reform published a study assessing Milwaukee’s School Choice Demonstration Project. For many voucher enthusiasts, the results were sobering. Students enrolled in choice schools performed no better on reading and math tests than students attending conventional public schools. Critics such as Kevin Carey, a leading center-left education reformer, suggested that the Milwaukee experiment is therefore a failure.

Yet as Frederick Hess of the American Enterprise Institute (AEI) noted, students enrolled in the choice program are educated more cheaply than district-school students. As of the 2008–09 school year, the maximum amount Milwaukee’s choice schools received per student was $6,607. In contrast, the Milwaukee public schools spent approximately $14,520 per pupil. While these numbers aren’t directly comparable, it certainly seems that the district schools are considerably more expensive than the voucher program. Moreover, parents were more satisfied with the quality of education at choice schools, which suggests that some of the schools’ positive aspects were not captured in reading and math scores.

In 2006–07, we spent $562 billion on K–12 public education, or 3.9 percent of GDP. Let’s be generous to the education industry and assume that we could trim its costs by only one-fifth. This would save taxpayers over $100 billion.

So far, the argument for school choice has been about equity and, for many suburban parents, the threat vouchers might pose to their property values. If it were framed as a question of value for money, and of restraining increases in the state and local tax burden, voters might find it considerably more attractive.

Targeted summer programs. Rather than simply cut $100 billion of educational spending, we could reallocate it in a variety of ways. Educational researchers Dave Marcotte and Benjamin Hansen found that ten additional instructional days had a bigger positive effect on a student’s educational outcomes than did a smaller class (the productivity-reducing go-to strategy of the teachers’ unions), repeating a grade, or even having a better teacher. And instead of extending the school year for everyone, we could do it on a voluntary basis, and for only the students who will benefit most.

In 2006, Princeton economist Alan Krueger and grad student Molly Fifer McIntosh published a paper calling for the creation of “summer opportunity scholarships,” vouchers designed to help students close the skills gap that opens when schools are out of session. During the school year, all students make educational gains. But students from disadvantaged families tend to lose ground over the summer months, while students from advantaged families, thanks to exposure to a more stimulating set of activities, lose very little, and might even make gains. Krueger calls this the Harry Potter divide: the gap between families that encourage recreational reading and other skill-building activities, and those that don’t.

Krueger and McIntosh suggest that summer opportunity scholarships be made available to all students eligible for free or reduced-price lunches. The funds could be redeemed only at carefully vetted institutions, operated by school districts, nonprofits, or businesses, that use tested and proven instruction techniques. They estimate that roughly 3.6 million students would take part at a cost of under $5 billion, a trivial sum in light of the program’s potential.

Discipline. In 2001, years after California spent large sums reducing class sizes, Stanford economist Edward Lazear published a paper finding that the number of disruptive students, to a greater degree than the total number of students, was a barrier to effective teaching. The misbehavior of one or two students demands the attention of the teacher, distracts from instruction for the rest of the class, and can be contagious. And even if we leave aside misbehavior, the presence of students who can’t keep up with demanding material and thus seek, and deserve, more personalized attention can disrupt the learning environment for other students. Very slight changes in student behavior can cause great differences in the amount of time spent on a task.

One obvious strategy would be to remove disruptive students from conventional classes and place them in a setting where they would receive the specialized attention they need, including more effective discipline. One of the reasons Milwaukee’s choice schools outperform district schools could be that teachers are willing to receive lower compensation if they know they won’t have to do constant battle with disruptive students.

Student-centric technology. The most promising avenue for delivering real productivity increases over time involves facilitating the spread of student-centric technology tools such as those described in Disrupting Class, written by celebrated management theorists Clayton Christensen, Michael Horn, and Curtis Johnson. Instead of focusing on the deployment of technology in existing classrooms, the authors call for using customized online educational tools to create courses that aren’t already offered. Even the largest high schools tend not to offer languages like Mandarin — or, for that matter, advanced math courses — and not everyone has access to a high-quality community college.

Online courses are already helping to address this problem, through programs such as the Florida Virtual School and Alabama’s ACCESS, which offer a rich complement of Advanced Placement and other courses to students attending conventional public high schools that lack them. Defenders of the status quo often argue that online courses are inferior to those conducted in a conventional classroom setting, but the objection is irrelevant if the choice is between a less-than-ideal way to learn Mandarin and no way to learn Mandarin at all.

And because online education remains an innovative, lightly regulated space, it is getting better and cheaper at a far faster rate than conventional alternatives. Which is to say that well-designed online tools could be the cure for Baumol’s cost disease in education.

Ultimately, all of these strategies emphasize increasing specialization in the education sector. Our schools are built on a model from the industrial age, when the operating assumption was that differences in learning styles were irrelevant — everyone had to master the same curriculum, so they would do it together. We have long since run up against the limits of that strategy. It’s very difficult to imagine that a single school will ever be able to offer the world’s best math instruction and Mandarin instruction, and also have the best record at mainstreaming autistic children. Student-centric technology can help us get closer to that utopian place by encouraging the rise of specialized institutions that focus on being the best at a fairly narrow task.

In his brilliant book Education Unbound, AEI’s Hess notes that conventional public-school districts tend to perform many functions in-house that comparable institutions in the private sector would outsource. Teacher recruitment, for example, is a function that a small, specialized firm could perform at lower cost and at higher quality than the in-house recruiting arms of most large districts. Also, in many states, the anti-reform movement is heavily subsidized by building contractors who depend on lucrative contracts to build and repair large, centralized facilities. A more diverse education marketplace would likely see the creative repurposing of commercial space, not to mention heavier reliance on online education — which would devastate the capital budgets on which politically powerful contractors depend.

The education marketplace is almost as dysfunctional as the health-care marketplace. As in health care, it is very difficult for consumers to get a good sense of the relative quality of different providers. Innovation has been severely restricted by the dominance of the government, and entrepreneurship has been relegated to the margins. That has to change.