School Choice Raises Student Achievement

Milwaukee's public schools raised their productivity quickly and dramatically in response to competition and ... the Milwaukee schools that faced the most competition raised their productivity the most.

In School Choice and School Productivity (Or, Could School Choice be a Tide That Lifts All Boats?)(NBER Working Paper No. 8873) author Caroline Hoxby calculates that average school productivity -- that is, student achievement per dollar spent -- was about 65 percent higher in 1970-1 than in 1998-9. If U.S. schools were only able to return to their 1970-1 level of productivity, the average U.S. student would be "scoring at an advanced level where fewer than ten percent of students now score," she writes. The dramatic decline in the productivity of American schools frustrates school reforms and imperils the traditional source of U.S. comparative advantage: skilled labor that supports advanced industries.

In this paper, Hoxby asks why school productivity is so low and whether increased school choice (the ability to choose between public schools within an area, or between public and private schools) might raise it. The question is an important one because many debates on school choice focus on students who might be "winners" and "losers." But, if school choice raises the productivity of schools, it will "lift all boats": that is, improve schools for all students. Recent productivity losses of American schools are so large that regaining a mere fraction (as little as a quarter) of recently lost productivity would "lift all boats."

Using scores from the National Assessment of Educational Progress (NAEP) as the measure of output (achievement) and real per-pupil spending on elementary and secondary education as the input, she examines the school productivity collapse. Hoxby considers whether changing family backgrounds can explain the collapse in schools' productivity. She finds that they cannot: while minority students now constitute a larger fraction of the U.S. student population and they tend to score worse on the NAEP, this phenomenon is overwhelmed by the fact that current students have parents who are better educated and have higher family incomes.

She also considers whether schools have faced rising costs because wages for females in professional jobs that are alternatives to teaching have been rising. She finds that, even if schools had had to give teachers the same wage increases as the most elite female professionals (medical doctors, lawyers) in order to keep teaching quality constant, the decline in school productivity would still be about 55 percent since 1970-1. In other words, Hoxby concludes, "School conduct, and not student characteristics or female career opportunities, is the main source of the decline in school productivity."

Hoxby examines the effect of choice on school productivity by looking at three recent reforms that have introduced choice into areas that previously had little: vouchers in Milwaukee, charter schools in Michigan, and charter schools in Arizona. She looks at the productivity of public schools that faced increased competition as a result of these reforms, not just at the productivity of the voucher or charter schools themselves. For instance, she compares the productivity of Milwaukee's public schools before and after the voucher program provided competition. As a control group for these schools, she uses urban public schools in Wisconsin that are located outside Milwaukee (and are thus immune from voucher competition) but that serve students similar to those of Milwaukee. She finds that Milwaukee's public schools raised their productivity quickly and dramatically in response to competition and that the Milwaukee schools that faced the most competition raised their productivity the most. Productivity rose because the schools achieved more while spending the same amount (as opposed to holding achievement steady while reducing spending). In fact, in the Milwaukee schools facing substantial competition, achievement rose by as much as 4.7 national percentile points faster per year than in control schools. Such gains are virtually unprecedented for an American school reform.

For Michigan and Arizona, Hoxby finds that even a very modest amount of charter school competition (the possibility of losing 6 percent of their students) makes public schools raises their productivity by a statistically significant amount. Greater charter school competition raises productivity even more. As with Milwaukee's voucher competition, the charter school competition impels productivity to rise through achievement gains at a steady level of spending.

Hoxby also reviews evidence on the effects of competition among traditional public school districts, noting that metropolitan areas with maximum school choice between school districts have "eighth grade reading scores that are 3.8 percentile points higher, tenth grade math scores that are 3.1 national percentile points higher, and twelfth grade reading scores that are 5.8 national percentile points higher." The productivity gains are more impressive that these achievement scores indicate because per pupil spending is also 7.6 percent lower in such metropolitan areas.

Finally, Hoxby reviews evidence that indicates that private school competition raises public school productivity through achievement gains (at a steady level of per-pupil spending). For instance: "A public school in a metropolitan area with moderately high [relative level of] private school choice has eighth grade reading scores that are 2.7 national percentile points higher, eighth grade math scores that are 2.5 national percentile points higher, twelfth grade reading scores that are 3.4 national percentile points higher, and twelfth grade math scores that are 3.7 national percentile points higher."

Hoxby concludes with some simple calculations that demonstrate that the productivity gains from school choice could easily swamp any other effects of choice. For instance, let's say that a Milwaukee student started with the best available peers in that city and (as a result of choice) ended up with the worst available peers. Let's also say that peers had an extraordinarily strong effect so that the student's achievement fell one-for-one with that of his peer group, as it deteriorated. Even in this extraordinarily pessimistic scenario, the effect of choice on Milwaukee productivity is so great that the student would be better off after only four years under the voucher reform.

-- Linda Gorman