Effects of high-achieving peers : findings from a national high school assignment system

Recent studies of US elite exam schools have yielded the startling conclusion that such schools improve neither educational achievement nor longer-term educational outcomes. Is the same true for exam schools elsewhere? The system in Turkey is ideal for investigating this question. There, students are placed in exam schools based on a high-stakes national examination. Utilizing an exceptional database for Turkey not heretofore available, we conduct regression discontinuity analysis exploiting score discontinuities between more than 200 exam schools. We find that attending more selective exam schools yields large achievement gains and improved university placements for high achieving students.