Some voucher skeptics argue that even if school vouchers benefit recipients, they do so by improving their peer groups at the expense of others’, and if so, there may be no net benefit to society as a whole. A necessary condition for this argument is that voucher recipients have more desirable peers than they otherwise would have. We take advantage of an educational voucher program in Colombia, for which spots were allocated by lottery, to identify a set of applicants for whom winning the voucher did not lead to attending schools with peers with superior observable characteristics. In particular, we focus on those who applied to vocational private schools. In this population, lottery losers rather than winners were more likely to attend academic secondary schools. Despite this, we find that even in this population, lottery winners had better educational outcomes, including higher graduation rates and reading test scores. This casts doubt on the argument that voucher effects operate entirely through improving the set of peers available to recipients. One hypothesis is that private vocational schools are much better than public ones at adjusting to the demands of the labor market. Consistent with this hypothesis, private vocational schools are overwhelming concentrated in teaching skills preparing students for Colombia's rapidly growing service sector whereas public vocational schools are much more likely to teach industrial curricula which prepare students for more traditional blue-collar positions.