This paper investigates the peculiarities that arise when mechanism design is deployed in contexts in which issues of social, racial and distributive justice are particularly salient. Economists’ involvement in redesigning Boston’s algorithm for allocating K-12 students to public schools serves as an instructive case study. The paper draws on the distinction between ideal theory and non-ideal theory in political philosophy and the concept of performativity in economic sociology to argue that mechanism can enact elaborate ideal theories of justice. A normative gap thus emerges between the goals of the policymakers and the objectives of economic designs. As a result, mechanism design may obstruct stakeholders’ avenues for normative criticism of public policies, and serve as a technology of depoliticization.
Thread from frequent co-author E. Glen Weyl here and embedded below:
I am blown away by the brilliant, award-winning paper by @zhitzig available here: https://t.co/T5tEiAJUKP. Everyone in market design needs to read it, as it absolutely nails what is so absurd about the whole Roth-style approach to matching market design.— (((E. Glen Weyl))) stands with 🇺🇦 and 🇹🇼 (@glenweyl) July 1, 2019
This paper follows in the footsteps of Alexandrova (2008) in critiquing the design of real-life markets using microeconomic theory. Whereas Alexandrova focused on game theory and the celebrated FCC spectrum auctions, Hitzig focuses on Roth-style matching theory and the allocation of applicants to Boston Public Schools (BPS), a canonical case study in the literature.
The titular “normative gap” refers to the inexactness with which the mechanism designed by the consulting economists actually applies their favored theory of justice, which Hitzig identifies as Arneson’s ’equal opportunity for welfare’ principle , a more complex normative framework than the notion of simple efficiency on which microeconomic theory usually hangs its hat. More precisely, Hitzig argues that the deferred acceptance algorithm ultimately advanced by the consultants could not be described as an application of a particular matching theory, but rather as the imposition of an idealized setting that coerces real people to role-play as the highly idealized agents the setting demands by assumption. Hitzig describes this as an enactment (as opposed to an application) arising from two idealizing assumptions: strict compliance (e.g. applicants are forced to disclose strict preferences they may not have) and favorable circumstances (e.g. applicants are forced to communicate complete preferences even if they do not have equal access to information on all schools). The language of coercion I use here derives from the implied consequence of non-participation being total exclusion from the BPS, the severity of which is dependent on the properties of the society external to the idealized setting. For the post-Brown v. Board of Education America in which the BPS mechanism was re-designed, this of course entailed a substantial degree of public hostility and systemic racism, an environment recently revisited on a national platform when the legacy of busing policy arose in the Democratic primary debates. Further, those excluded from the idealized system are typically also excluded from the market’s welfare calculus.
Hitzig argues that such coerced enactment gives rise to “pernicious” outcomes in some circumstances: “when ideal theory is enacted as in the BPS case, it might facilitate the evasion of a proper discussion of the real injustice particular to a given situation”, in this case that of racial inequality and “domination.” I am readily sympathetic to this perspective, but feel this argument would greatly benefit from more directly linking the normative gap to the mechanism’s failing marginalized groups. To me, marginalization does not clearly follow from the imprecision of stated preferences, for example. The Weyl thread linked to above infers severe implications, suggesting the argument conveys that Roth-style matching theory
[hides the flaws of the underlying logic of capitalism] behind a pretense of satisfying abstract properties in an extremely narrow setting while neglecting e.g. that some participants are desperately poor and abused and that their outside option is jail while others can go to prep school"
and that it
[uses] formalism to kill off such contentious debates about neoliberalism…destroying in a haze of technical purity our ability to honestly debate the way capitalism and resulting monopolies are ravaging our society
Perhaps owing to my need to have things spelled out, this didn’t neatly follow from my reading of the paper (or at least the most recent draft I have access to) since it makes no reference to specific real-life implications of the normative gap other than vaguely as a distraction from addressing systemic social problems. The paper at is at multiple points purposefully framed away from a rejection of the redesigned BPS mechanism either in theory and practice. There are references to the shortcomings of matching theory’s underlying dependence on rational choice and opportunities for behavioral economics to act as a cautionary force on these technocratic prescriptions in order to accomplish the titular ‘bridging’, but I’m not clear on what that could entail.
Nonetheless, it was a refreshing critical complement to my recent matching-markets module. The list of things people have argued would improve traditional PhD economics training is ever-growing and includes instruction on the history of economic thought, the demotion of macroeconomics from a core course to a field course, and guidance on the actual practice of research. Integration of critical perspectives such as these is another.
Graduate economics students will be familiar with the correspondence of different market designs to different theories of justice. A welfare module might entail comparing the graphical and algebraic representations of utilitarian, Rawlsian, and dictatorial social welfare functionals. Any treatment of matching markets will formalize the tradeoffs among the properties of Pareto efficiency, strategy-proofness, and stability (and all sub-variants thereof). But my feeling is that we are maybe susceptible to the appeal of “mechanical objectivity” that Hitzig argues is inherent to mechanism design beyond just the familiar inadequacies of Pareto.
Hardly original to to see Donald Trump’s comedic immunity as also applicable to Boris Johnson. Still, I’m thinking of Bo Burnham’s take Donald Trump’s comedic immunity as also applicable to Boris Johnson:
I personally felt like Trump is a product of comedy as much as anything. I mean, when I watched him in those primaries, like, I could see the tactics that he used from the Comedy Central roast that he was a part of. I can absolutely see he used the tactics of comedic timing and ribbing and all of that. And when I watched the primaries and watched him dismiss everybody and watched him make fun of everybody, I’m like he is winning this in the way a comedian wins a room.
It was so funny to then watch the comedy community scramble around and act like we are the one thing that can take him down when it’s, like, are you kidding me? Like, this is the exact thing that’s being used for him. How are you going to make fun of this guy? How are you going to satirize this guy who is using, you know, laughter to set the world on fire?
A European official on Johnson:
We answer his attacks, but the problem is that our answers are not funny.
Go Tell It On The Mountain, James Baldwin (h/t Junho and Sam)
Relatable depiction of a young protagonist’s guilt and shame attempting to reconcile the Christian doctrine with the process of coming of age. The novel’s shifting perspectives through generations of a family immediately recalled East of Eden, one of my all-time favorite books. The comparison carries through to its enigmatic ending. Was curious but couldn’t find anything online on how the timing of the novel’s writing tracks the evolution of Baldwin’s own personal relationship with his faith.