Saturday, August 31, 2019

Predicting stable matches from the preferences of one side of the market: Haeringer and Iehlé in AEJ-Micro

Two-Sided Matching with (Almost) One-Sided Preferences
By Guillaume Haeringer and Vincent Iehlé
American Economic Journal: Microeconomics 2019, 11(3): 155–190.


Abstract: "In a two-sided matching context we show how we can predict  stable matchings  by  considering  only  one  side’s  preferences  and  the  mutually  acceptable  pairs  of  agents.  Our  methodology  consists  of  identifying  impossible  matches,  i.e.,  pairs  of  agents  that  can  never  be matched together in a stable matching of any problem consistent with  the  partial  data.  We  analyze  data  from  the  French  academic  job  market  for  mathematicians  and  show  that  the  match  of  about  45 percent of positions (and about 60 percent of candidates) does not depend on the preferences of the hired candidates, unobserved and submitted at the final stage of the market."


Haeringer and Iehlé present new theory and explore an interesting data set, described as follows:

"Market for Mathematicians
In 1998, a small group of young mathematicians set up a website, Opération Postes,  inviting  recruiting  committees  to  announce  the  lists  of  candidates  to  be  interviewed  as  well  as  the  rankings  of  candidates  that  will  be  submitted  to  the  clearinghouse (the ministry), as soon as these would be decided.19 The community of mathematicians was very responsive and the website quickly became a central tool  in  the  job  market.20  The  data  for  each  position  (interviewees  list  and  rank-ings) is usually uploaded by the the chairs of the recruiting committees themselves (and  if  not,  by  a  member  of  the  committee).  On  average,  about  90–95 percent  of  the  job  openings’  interview  lists  and  rankings  are  available.21  The  data  of  Opération  Postes  is  public,  although  not  in  a  format  that  makes  it  immediately  usable  for  any  analysis.  There  are  many  misspellings,  and  we  sometimes  found  confusions  between  the  married  and  maiden  names  of  some  female  candidates.  By  cross-referencing the data with other sources we were able to compose a clean dataset.22We  also  collected  for  each  year  the  assignment  of  candidates  to  departments.  This  assignment  is  computed  by  the  Ministry  of  Higher  Education  by  using  candidate’s  submitted  preference  lists  over  the  departments  and  the  rankings  of  candidates established by the recruiting committees."

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.