From left, Vice President-elect Biden, President-elect Obama and Arne Duncan, then chief executive officer of the Chicago public school system, speak to students at Dodge Renaissance Academy in Chicago in 2008. Duncan would become education secretary in the Obama administration. (Ralf-Finn Hestoft/Pool via Bloomberg News)

In 2005, the school board in Kalamazoo, Mich., unveiled a radical offer that's since been celebrated, studied and mimicked across the United States: Anonymous donors there had pledged -- indefinitely -- to help pay for in-state college tuition for any child who came up through the city's public schools.

The idea, the Kalamazoo Promise, took a strategy that had been tried on a much smaller level and scaled it up to an entire urban school system. Evidence collected over nearly a decade suggests that the "promise" helped lead to higher enrollment in the city's public schools, improved academic achievement, better graduation rates and boosted college enrollment. But a new national study of Kalamzoo's copycats presents a more complicated picture of what happens when this idea spreads elsewhere.

University of Pittsburgh economists Michael LeGower and Randall Walsh found some national patterns that are both promising and unexpected: In nearly two-dozen cities and towns that have tried some variation on the promise, public school enrollment promptly increased, as did housing prices in the neighborhoods around many schools.

But these promised scholarships -- for now -- are having the largest impact on middle- and upper-income households, not lower-income ones. And the design of some scholarship programs may actually exacerbate racial achievement gaps in education.

LeGower and Walsh's findings, published in a National Bureau of Economic research working paper, suggest that above all, the details of the promise matter.

The original Kalamazoo program was meant to boost the prospects of children already in the city's public schools. But business leaders hoped that the ripple effects would be even broader, transforming the school system itself, and the city's economic development along with it. Offer quality public education, the thinking went, and upper-income parents -- and the kinds of companies that want to employ them -- will move to town and stay there, too.

Kalamazoo offers scholarships at any public college or university in Michigan to students who have lived in the city and attended its public schools through high school. The scholarship is designed on a sliding scale depending on how long children have been in the school system, with those starting in kindergarten getting the largest reward upon high school graduation: a full four-year scholarship, including tuition and fees.

The Kalamazoo program has no grade-point average cutoffs or merit requirements. Any student who graduates from the public-school system after at least four years qualifies. But many "promise" programs elsewhere around the country are merit-based. Or they offer scholarships to a narrower range of schools, sometimes at a much smaller scale. The program in Jackson County, Mich., for example, promises a maximum of just $600 for two years.

The differences between all of these programs enable LeGower and Walsh to identify which kinds of promises have the greatest impact. And because these scholarship gifts -- often from the business community or philanthropists -- are typically unexpected, their announcement creates a series of natural experiments. How did school enrollment and housing prices change, for instance, immediately after these gifts were announced?

It's impossible to study the true counter-factual: What would happen in these communities if these scholarships were never offered? But by comparing enrollment data for surrounding schools (without scholarships), LeGower and Walsh identify an immediate uptick in enrollment in the public school systems where scholarships were offered:


Across 22 programs, including Kalamazoo's, LeGower and Walsh find an increase in total public school enrollment of about 4 percent in the years immediately after the announcement. Most of this gain came from children enrolling in kindergarten through fourth grade, when we would expect parents to try to take advantage of the maximum scholarship allotment. Not surprisingly, programs offering scholarships to all students regardless of merit, and to the widest range of colleges and universities, saw the biggest gains in enrollment, of about 8 percent.

Merit-based scholarships, on the other hand, prompted a particular boost in white enrollment. But they also appear to have led to decreases in non-white enrollment.

LeGower and Walsh also looked at property values within a subset of eight cities where housing data was available. Within three years of a promise announcement there, they found, housing prices in the neighborhoods around the best-performing public schools increased by 7 to 12 percent (or as much as about $20,500). This suggests that families who could afford to were paying a premium to access the scholarships.

Why would that matter for the overall impact of these programs? "This could have the effect," LeGower and Walsh write, "of contributing to further inequality in educational outcomes if the high-income households attracted by Promise programs are exclusively attending already high-quality schools."

For many families, these scholarships close the gap between what they can pay and what higher education actually costs. And so families already inclined to send their children to college are likely to see the greatest value in them. Merit-based scholarships also primarily benefit students with the highest achievement, attending the schools that already offer merit-based programs like Advanced Placement classes.

If a limited scholarship program draws new upper-income families into only these schools, while providing little benefit to existing students in struggling public schools, such a "promise" could theoretically stretch the achievement gap even further. These new upper-income families might bring with them other benefits, like higher tax revenue for the entire school system. But if one of the goals of a "promise" program is to bring new middle- and upper-income students into the same classrooms as lower-income peers, some of these programs likely may not have that effect.

That result underscores the tremendous challenge of improving public schools for the children already in them, even as cities try to lure upper-income families back into public education. A "promise" program that serves all of them -- offering universal scholarships, and to a range of schools that would benefit both National Merit scholars and first-generation graduates -- could be the best path there.