Throw Me the Money

Mayor Bill Purcell's proposed $1.45 billion budget for the city would do a lot of things for a number of city services and departments. But it's the spending increase for education that always seems to draw the most attention, especially when tax hikes are part of the plan. With Purcell proposing both property and sales tax hikes, the enduring question of a connection between financial resources and school quality is sure to follow.

Does more money translate into better schools? The anti-tax crowd has been crying "no" for years. Many educators and liberals have been replying "does too" for years. Does it? Let's try to sort it out. A warning up front: no matter which side of the argument you're on, you may not like the answer.

The starting point is 1981, when economist Eric Hanushek reviewed numerous studies of how schools turn "inputs" (including money) into "outputs" (measures of educational performance). In an article titled "Throwing Money at Schools," he concluded with vigor that "there is no relationship between expenditures and the achievement of students." The one thing that did have an effect was teacher experience, although Hanushek wondered if the cause-effect link on that might run the other way: quality teachers choosing to work at better schools.

Hanushek's work sparked a generation of studies and arguments about whether it could really be true that expenditures make no difference. It also sparked a durable cottage industry in use of the shibboleth "throwing money at schools" to mock educational policy. Googling that phrase turns up more than 750 hits, many of them conservative pundits, analysts and bloggers bemoaning decades of higher spending and stagnant performance. As a Heritage Foundation report declared in 1999, "money does not cure the problems ailing America's schools."

But this certainly didn't settle the issue—there are simply too many who feel in their gut that money has to matter. How can it not, after all? Money buys things that schools need, from better teachers to better facilities, textbooks and equipment. If money doesn't matter, then why do the best performing public schools tend to be in relatively affluent neighborhoods with larger PTO budgets? If money doesn't matter, then why do the toniest private schools charge tuition that far outpaces per-pupil spending in the public schools? If money doesn't matter, then why are the most sought-after private schools for college-bound students the ones with the fanciest digs charging the highest tuition?

Hanushek updated his work in an important 1997 paper that looked at 59 studies gathering together almost 300 estimates of the link between resources and educational quality. The result was similar: he reported a positive link in fewer than one in six estimates. Present-day conservatives love to cite Hanushek's work to back up their anti-"throwing" position.

But wait: enter from stage left Princeton economist Alan Krueger, who challenged Hanushek's results in a 2003 paper that took issue with Hanushek's methods. (I'll spare you the details.) Krueger reanalyzed the data and concluded that "resources are systematically related to student achievement." So it turns out it's not just the wingtips and the Birkenstocks who are yelling "does too" and "does not" back and forth; the respected social scientists are doing it too.

By now you're ready for the paragraph that sorts out the truth. But there are two problems with the research that get in the way. (Actually there are several, but let's keep it to two.) First, studying money's role in education is a kind of a rigged game from the start because it's so much easier to find no connection than to find a link—not because the link isn't there, but because it's difficult to study effectively. Here why. The let's-not-throw crowd likes to mention long-term trends in American education going back to the 1970s: decades of higher spending and flagging test scores. True enough by the numbers, but it's risky to draw serious conclusions from them without taking into account all (or at least some) of the other potentially important factors that twist and turn in the messy process of public education. It's especially dicey when looking at spending and test scores in a large, aggregated way—a whole city school system, or a state, or the entire country.

Given so many individual schools with diverse populations, existing resources, strengths and flaws, it's hard to believe you'd ever find a clear link between changes in a single input (dollars) and changes in a single measured outcome (such as a test score). When you don't find the link, you can't really say that money has no effect if you've ignored other key factors that interfere with that outcome.

Let me illustrate with a hypothetical example. Imagine a school system earmarks new money to improve its high school dropout rate. Assume that the kids who would have dropped out now stay in school, but do poorly on standardized tests, while some other students' scores improve. You'd have a situation in which a rise in spending accomplished its goals, yet the statistical link between change in dollars and change in test scores nets out to zip.

This is not a fatal problem for researchers: you can study the effects of money on performance—if you can find specific, measurable things to spend the money on so that we can see their effects (and if you control for the right "other things" that might contaminate the link). Researchers have done this, which brings us to problem number two. The research has focused almost exclusively on one particular educational investment that happens to be easy to quantify, measure and compare: class size.