Research

Disparate Treatment and Outcomes in Emergency Departments: Evidence from Florida (with Ozde Ozkaya)
Job Market Paper

Strokes are roughly twice more likely to be missed among Black patients, with most of the disparity arising from physicians testing Black patients less often. We develop a method to quantify the role of disparate treatment by physicians in driving this difference in testing. Specifically, we leverage a unique feature of strokes: whether a patient actually had a stroke can be inferred retrospectively even if initially misdiagnosed. This allows us to benchmark testing decisions against racially objective predictions of stroke risk made by a machine learning model trained on the true underlying stroke states. We decompose disparate treatment into two forces: an unjustified skill gap, where physicians make noisier risk assessments for Black patients; and racial prejudice, where physicians are less likely to test Black patients conditional on their risk assessment. Disparate treatment accounts for about 65% of the racial disparity in testing. Removing racial prejudice would lower testing disparities by half.

Who is Sex-Selecting, and When?

This paper studies fertility choices of parents in India who typically have a strong preferential bias for sons over daughters. Such parents are known to engineer the sex-composition of their children using prenatal sex-selective abortions or continued childbearing. I propose and empirically validate a general heuristic or a rule that describes when parents decide to sex-select. The decision to sex-select is motivated by a desire for more sons and necessitated by a high cost of continued childbearing. Identifying a heuristic of sex-selection therefore requires: (1) accounting for parental preferences over sex-composition and total children, and (2) observing parents' decisions of sex-selection. Towards the first, I use data on mothers' ideal number of children (self-reported) to define relative birth orders for all births in my sample indicating how far they are from their mother's ideal number of children. The use of sex-selection at any relative birth order is indicated by skewed male-to-female births at the order. I find the ratio of male-to-female births to be the highest for children born at their mother's ideal number of children. This suggests a heuristic whereby parents sex-select when at their ideal number of children, to avoid exceeding it. I empirically validate this heuristic by exploiting the natural orthogonality between sex assigned at birth and the birth interval that precedes it. This orthogonality breaks down only if sex-selective abortion is resorted to, i.e. birth intervals preceding male births become artificially longer than those preceding female births. Following birth histories with no or few male births, I find intervals preceding male births to be longer for parents who are at their ideal parity than others—as the heuristic suggests.

Group-Bias in Interpersonal Interactions

Groups, including those created based on trivial and meaningless criteria, exhibit group-bias by favoring their in-group relative to the out-group. What generates this bias has been a source of debate: Is it an effect of salience in categorization, or is what appears to be group-bias merely a consequence of strategic behavior to gain from the interdependence of payoffs in settings with groups? Using groups induced in the lab, I experimentally manipulate payoff structures to study what drives this bias in interpersonal interactions. I find subjects to favor their assigned in-group even when their respective in-groups cannot affect their payoffs. Categorization is hence a sufficient source of group-bias that operates even in the absence of any strategic pecuniary interests or expectations of generalized reciprocity. Additionally, I find subjects to attach importance to group labels more than just as a salient dimension of categorization, and care about how they’re perceived by their in-group. Subjects are asked to nominate two out of a total of four group members for a bonus, and the nomination results are revealed only to treated groups. Subjects informed of being nominated by their in-group not only increase their allocations to the in-group but also decrease allocations to the out-group, thereby exhibiting more group-bias. This is consistent with subjects identifying with their assigned group and drawing non-pecuniary gains from the categorization. Those informed of their non-nomination decrease allocations to all and do not maintain any positive distinctness between their in-group and out-group, as if having disidentified from the group labels.

Minu Philip

Working Papers

Works in Progress

Publications