Objective. In 1975, the Puerto Rican community successfully sued the New York City Department of Education, mandating the city to provide bilingual education to its Spanish‐speaking English learner (EL) students. The settlement, known as the “Aspira Consent Decree,” has been amended over time to include EL students of all language groups and now requires public schools that have at least 15 students of the same language group in two contiguous grades to offer bilingual education. Yet observational studies of bilingual education classrooms in the city document that Spanish‐speaking EL students may be the only language group that receives native‐language instruction, while students from other language groups who are enrolled in bilingual education primarily receive English instruction.
Method. I use this difference in treatment dosage to estimate the effect of bilingual education on the time that it takes students to learn English.
Results and Conclusions. Students who enroll in bilingual education classrooms learn English less quickly, but the effect of bilingual education is the same for Spanish‐speaking and other students, implying that negative selection mechanisms are at work.