I heard Yusuke Narita present this remarkable paper comparing centralized and decentralized school matching at a recent seminar at Stanford. It uses a data set involving both outcomes and rules (rules are data!) of a school choice system for elite schools in Japan from the turn of the last century
Meritocracy and Its Discontent: Long-run Effects of Repeated School Admission Reforms
TANAKA, Mari NARITA, Yusuke MORIGUCHI, Chiaki
Abstract: "We study the impacts of changing school admissions systems in higher education. To do so, we take advantage of the world’s first known implementation of nationally centralized admissions and its subsequent reversals in early twentieth-century Japan. This centralization was designed to make admissions more meritocratic, but we find that meritocracy came at the cost of threatening equal regional access to higher education and career advancement. Specifically, in the short run, the meritocratic centralization led students to make more inter-regional and risk-taking applications. As high ability students were located disproportionately in urban areas, however, increased regional mobility caused urban applicants to supplant rural applicants from higher education. Moreover, these impacts were persistent: four decades later, compared to the decentralized system, the centralized system continued to increase the number of urban-born elites (e.g., top income earners) relative to rural-born ones.
" Our empirical setting is the first known transition from decentralized to nationally centralized school admissions. At the end of the 19th century, to modernize its higher education system, the Japanese government set up elite national schools (high schools or colleges) that served as an exclusive entry point to the most prestigious tertiary education (Yoshino, 2001a,b; Takeuchi, 2011). These schools later produced many of the most influential members of the society, including several Prime Ministers, Nobel Laureates, and founders of global companies like Toyota. Acceptance into these schools was merit-based, using annual entrance examinations. Initially, the government let each school run its own exam and admissions based on exam scores, similar to many of today’s decentralized K-12 and college admissions. The schools typically held exams on the same day so that each applicant could apply for only one school. Similar restrictions on the number of applications exist today in the college admission systems of Italy, Japan, Nigeria, and the UK.
"At the turn of the 20th century, the government introduced a centralized system in order to improve the quality of incoming students. In the new system, applicants were allowed to rank multiple schools in the order of their preference and take a single unified exam.1 Given their preferences and exam scores, each applicant is assigned to a school (or none if unsuccessful) based on a computational algorithm. The algorithm was a mix of the so called Immediate Acceptance (Boston) algorithm and Deferred Acceptance algorithm with a meritocracy principle imposed upfront. To the best of our knowledge, this instance is the first recorded, nation-wide use of any matching algorithm. Furthermore, for reasons detailed below, the government later re-decentralized and re-centralized the system several times, producing multiple natural experiments for studying the consequences of the different systems."
Meritocracy and Its Discontent: Long-run Effects of Repeated School Admission Reforms
TANAKA, Mari NARITA, Yusuke MORIGUCHI, Chiaki
Abstract: "We study the impacts of changing school admissions systems in higher education. To do so, we take advantage of the world’s first known implementation of nationally centralized admissions and its subsequent reversals in early twentieth-century Japan. This centralization was designed to make admissions more meritocratic, but we find that meritocracy came at the cost of threatening equal regional access to higher education and career advancement. Specifically, in the short run, the meritocratic centralization led students to make more inter-regional and risk-taking applications. As high ability students were located disproportionately in urban areas, however, increased regional mobility caused urban applicants to supplant rural applicants from higher education. Moreover, these impacts were persistent: four decades later, compared to the decentralized system, the centralized system continued to increase the number of urban-born elites (e.g., top income earners) relative to rural-born ones.
" Our empirical setting is the first known transition from decentralized to nationally centralized school admissions. At the end of the 19th century, to modernize its higher education system, the Japanese government set up elite national schools (high schools or colleges) that served as an exclusive entry point to the most prestigious tertiary education (Yoshino, 2001a,b; Takeuchi, 2011). These schools later produced many of the most influential members of the society, including several Prime Ministers, Nobel Laureates, and founders of global companies like Toyota. Acceptance into these schools was merit-based, using annual entrance examinations. Initially, the government let each school run its own exam and admissions based on exam scores, similar to many of today’s decentralized K-12 and college admissions. The schools typically held exams on the same day so that each applicant could apply for only one school. Similar restrictions on the number of applications exist today in the college admission systems of Italy, Japan, Nigeria, and the UK.
"At the turn of the 20th century, the government introduced a centralized system in order to improve the quality of incoming students. In the new system, applicants were allowed to rank multiple schools in the order of their preference and take a single unified exam.1 Given their preferences and exam scores, each applicant is assigned to a school (or none if unsuccessful) based on a computational algorithm. The algorithm was a mix of the so called Immediate Acceptance (Boston) algorithm and Deferred Acceptance algorithm with a meritocracy principle imposed upfront. To the best of our knowledge, this instance is the first recorded, nation-wide use of any matching algorithm. Furthermore, for reasons detailed below, the government later re-decentralized and re-centralized the system several times, producing multiple natural experiments for studying the consequences of the different systems."
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.