Abortion and Mental Health - The role of Selection
-
Series
-
Speaker(s)Lena Janys (University of Bonn, Germany)
-
FieldEmpirical Microeconomics
-
LocationOnline
-
Date and time
March 02, 2021
16:00 - 17:00
Please send an email to Nadine Ketel or Paul Muller if you are interested to participate in this seminar (series).
Abstract:
Mental health consequences of abortions have recently been the subject of public discussions surrounding abortion legislation in several countries. Yet, it remains unclear whether the positive association commonly found in the literature reflects a causal effect or whether the main driver is selection. Using administrative records from both in- and outpatient data among a population of Swedish women aged 16-23 in the region of Skane, we show that there are substantial differences in the rate of diagnoses of mental health disorders and risky health behavior between women who experienced an abortion and those who did not; however, we find precisely estimated null-effects of abortions once we control for time-varying unobserved heterogeneity. To do this, we implement an estimator using group-fixed effects first proposed by Bonhomme and Manresa (2015) where unobservables are allowed to vary over time. Recognizing that even small differences in innate mental health might amplify over time, we formulate a possible mechanism for the starkly differing time-profiles in unobserved heterogeneity: we hypothesize that there is a dynamic interaction between risky health behaviors that increase the probability of an abortion and mental health. We develop a theoretical model that is consistent with the notion that initial conditions have a highly nonlinear effect on future mental health paths via present biased preferences that lead to more risky health behavior and, using the moments estimated from the data, we can reproduce the estimated observed time profiles. We validate our intuition that the initial selection is into risky behaviors that increase both mental health risks and unintended pregnancies we show that the different risk groups are equally associated with other types of risky (sexual) behavior. These findings have implications for policies aiming to restrict abortion access and for strategies trying to effectively screen for mental health conditions in adolescents.