A Structural Model of Charter School Choice and Academic Achievement

School Reform, May 2013

Lottery-based instru­men­tal vari­ables esti­mates show that Boston’s char­ter schools sub­stan­tially increase test scores and close racial achieve­ment gaps among their appli­cants. A key pol­icy ques­tion is whether char­ter expan­sion is likely to pro­duce sim­i­lar effects on a larger scale. This paper uses a struc­tural model of school choice and aca­d­e­mic achieve­ment to extrap­o­late from IV esti­mates and pre­dict the effects of char­ter expan­sion for the city­wide achieve­ment dis­tri­b­u­tion in Boston. Estimates of the model sug­gest that char­ter appli­cants are neg­a­tively selected on achieve­ment gains: low-income stu­dents and stu­dents with low prior achieve­ment gain the most from char­ter atten­dance, but are unlikely to apply to char­ter schools. This form of selec­tion implies that char­ter schools are likely to pro­duce sub­stan­tial gains for mar­ginal stu­dents drawn in by expan­sion. Simulations sug­gest that real­is­tic expan­sions are likely to reduce gaps in math scores between Boston and the rest of Massachusetts by 10 per­cent, and reduce racial achieve­ment gaps by 5 per­cent. Nevertheless, the esti­mates also imply that per­ceived appli­ca­tion costs are high and that most stu­dents pre­fer tra­di­tional pub­lic school to char­ter schools, so large expan­sions are likely to leave many char­ter seats empty. These results sug­gest that the poten­tial gains from char­ter expan­sion may be lim­ited as much by demand as by supply.