We propose an approach for credibly estimating indirect sorting effects of major education reforms and placing them alongside the reforms' direct and persistent effects for the first time. Applying our approach to California's statewide class size reduction program, we estimate a large positive direct effect of smaller classes on test scores and an even larger indirect effect due to demographic changes as private school students switch into public schools; both effects also persist. Accounting for sorting using these estimates raises the program's benefit-cost ratio significantly. Further, our analysis indicates that indirect sorting is likely relevant in policy evaluations more generally.