Jump to Navigation
Sabancı University
Faculty of Arts and Social Sciences

In Search of a School Placement System

Ahmet Alkan
(Sabancı University)
October 11, 2017, Wednesday
FASS 2034, 11:40
Once again, more drastically than before, the centralized system in Turkey for the placement of students into lycees and into universities is subject to criticism and change. The fact that students get ranked according mainly to their Exam scores lies at the center of the discussion :  It is maintained that the quality of preceding education suffers thereby, in particular by the multiple-choice nature of the Exam questions, and but alternative remedies raise issues of sound and fair assessment, of equal treatment vs nepotism, of assessment costs, of coordination vs chaos, and of equal opportunity, among others. 
The current theory of matching markets and practice in market design offers insights into some of the issues above, notably allocational efficiency and need for coordination, but not quite more than terminology in addressing the others. This talk will attempt to go over the issues within a common framework so as to view the main tradeoffs involved in institutional design. 
A particular look will be on letting schools assess students in decentralized manner. In this regard, some findings from “Matching with Information Acquisition” (2015) by  Z. H. Alioğulları and A. Alkan will be presented. This working paper considers matching under costly information acquisition (e.g., interviewing) in a stylized one-to-one market where (i) Students’ rankings over Universities are based on perfect information but (ii) Universities’ rankings over Students are based on partial information (e.g., exam score) and improvable by interview.  Each University in Stage 1 chooses two students to interview and in Stage 2 submits a ranking to a Center which implements the University optimal stable matching.  This game mimics the U.S. National Intern Resident Matching Program game. We  show that pure strategy Nash equilbria, implying stable interview schedules, may fail to exist. There is a unique pure strategy equilibrium where interviewees hierarchically overlap (forming a “ladder”) if and only if students’ preferences are “effectively homogeneous”.