Playing on a Level Field: Sincere and Sophisticated Players in the Boston Mechanism with a Coarse Priority Structure

Moshe Babaioff, Yannai A. Gonczarowski, Assaf Romm

Introduction

In the design of school choice systems, the choice of assignment algorithm is often framed as a trade-off between efficiency and fairness. Two mechanisms dominate the landscape: the Boston Mechanism (BM) and the Deferred Acceptance (DA) algorithm. Historically, the Boston Mechanism has been criticized for being "manipulable." Because BM prioritizes students based on their stated first choices, parents often feel pressured to strategize—sometimes ranking a less-preferred school first to ensure a seat, rather than "wasting" their top choice on a highly competitive school.

Conventional wisdom, largely shaped by the seminal work of Pathak and Sönmez (2008), suggests that this manipulability creates a systemic disadvantage for "sincere" parents (who report their true preferences) compared to "sophisticated" parents (who game the system). However, that research relied on the assumption of strict priorities, where every student has a unique rank. This paper challenges that narrative by examining markets with coarse priority structures, where many students share the same priority level (e.g., neighborhood residents) and ties are broken randomly. In these more realistic settings, the authors find that the "sophisticated advantage" often disappears, and sincerity can even become a strategic asset.

The Shift to Coarse Priorities

In real-world school districts, priorities are rarely unique. Schools typically group students into broad categories such as "siblings of current students" or "residents of the school zone." Within these categories, students are essentially indistinguishable to the algorithm until a random tie-breaker is applied.

This paper employs a "large random market" model to analyze how these weak priorities change the game. By moving away from strict rankings, the authors demonstrate that the outcomes of BM become highly sensitive to specific market configurations. This sensitivity leads to an "anything-goes" environment where no single group—sincere or sophisticated—is guaranteed to come out ahead.

The Sincerity Advantage: The Stackelberg Effect

One of the paper's most striking findings is the Sincerity Advantage. In a market with coarse priorities, a sincere student who ranks a popular school first effectively "commits" to that seat. Because sophisticated students are aware of this, they may observe the high demand from sincere players and rationally decide to "shade" their own demand, applying elsewhere to avoid the risk of a random tie-breaker they might lose.

In this dynamic, the sincere student acts as a Stackelberg leader. Their lack of flexibility forces the "smart" players to react and adjust their strategies. While sincerity can still be a disadvantage because the student cannot respond to excess demand elsewhere, the "crowding out" effect provides a powerful counter-balance that can level the playing field.

Reconciling Theory with Global Empirical Data

The theoretical framework provided in this paper helps explain why empirical studies of school choice have yielded such contradictory results across different cities:

  • Cambridge, MA & Barcelona: In these cities, sophisticated players were found to gain significantly more than sincere ones.
  • Beijing: This case stands out as the only known instance where sincere students showed relative gains over sophisticated ones.
  • Amsterdam: Here, students actually achieved higher expected utility under the strategy-proof DA than under BM.

The authors' "anything-goes" theorem (Theorem 3.1 and 4.2) proves that these variations are not anomalies. Depending on the utility levels of the schools and the ratio of sincere-to-sophisticated players in the population, any of these outcomes is theoretically possible.

Key Results and Implications

The paper provides several formal proofs that upend the classic understanding of school choice:

  1. Preference for Sincerity: In large random markets, there is always a constant fraction of students who would strictly prefer to be sincere rather than sophisticated.
  2. DA vs. BM: Contrary to previous theories suggesting sophisticated players always prefer BM, the authors prove (Proposition 4.1) that with weak priorities, a sophisticated student's utility in BM can be strictly lower than in DA. This is because BM effectively forces students to gamble on a single overdemanded school, whereas DA allows them to compete for multiple options.
  3. No Universal Winners: Theorem 4.2 establishes that in any large market, there will be groups of students (both sincere and sophisticated) who prefer DA, and groups who prefer BM.

Conclusion

By focusing on the coarse priority structures that define real-world school choice, this research suggests that the Boston Mechanism may not be as inherently biased against sincere parents as previously feared. The "level playing field" emerges not from the fairness of the mechanism itself, but from the strategic complexity introduced by random tie-breaking and the commitment power of sincerity. While the paper acknowledges limitations—such as the assumption of a single priority class—it provides a vital new perspective for policymakers: the choice between BM and DA is not a simple matter of protecting the unsophisticated, but a complex decision with no universal winner.

Summary updated

BibTeX
@inproceedings{babaioff2019playing,
  author = {Moshe Babaioff and Yannai A. Gonczarowski and Assaf Romm},
  title = {{Playing on a Level Field: Sincere and Sophisticated Players in the Boston Mechanism with a Coarse Priority Structure}},
  booktitle = {EC},
  year = {2019},
  doi = {10.1145/3328526.3329596},
}
Figure