EN
Back to the archive

The encyclopedia · People & Management · Operational decision · 2014

Israel replaced its intern lottery with a match that trades placement chances

Israel's medical intern assignment now trades placement probabilities between students, giving almost everyone a better shot at their preferred hospital.

Israel Ministry of Health

the move

Before 2014, Israeli medical interns were assigned by a random serial dictatorship: a lottery drew an order and students chose from remaining posts, in an auditorium with ID numbers drawn from a hat in earlier years.

A collaboration of medical professionals, economists and computer scientists redesigned the clearinghouse. The new mechanism runs the old lottery many times to estimate each student's probability of each hospital, then trades those probabilities to improve everyone's expected rank.

The redesign, used for the first time in 2014, produces significant efficiency gains: more students get a first or second choice, and no student is made worse off in probability terms. Researchers caution the scheme is no longer fully strategy-proof and needs monitoring.

why it works

  • Simulating the old lottery reveals inefficiencies that ex-post stability hides.
  • Probability trading lets students who prefer a hospital buy better chances from those who do not.
  • Rank efficiency improves the distribution of choices for nearly all students.
  • Centralised preference data also revealed which hospitals students actually prefer.
the payoffTrade probabilities to lift expected rankinspired

what transfers

When a market allocates by lottery, simulate it and trade the underlying probabilities: outcomes improve for almost everyone without hurting anyone.

what came after

The mechanism has run for successive intern cohorts and influenced later Israeli market-design work in psychology internships and other matching markets. It became a textbook case of applying mechanism design to a purely non-monetary allocation.

references

spotted an error? The archive wants to know.

same kind of clever