• Physician networks and patient referrals in Quebec:

With Vincent Boucher and Maripier Isabelle

Forthcoming Working Paper

This paper examines the impact of a centralized referral system on patient allocation among specialists and physician networks. We focus on the implementation of the "Centres de Répartition des Demandes de Services" (CRDS) in Québec, using unique administrative data on all services provided to patients by specialist physicians from 2011 to 2020. Our analysis consists of two stages: (i) we applied a microeconomic model and queuing theory to estimate the relationships between general practitioners and specialists based on their empirical networks; (ii) we employed "Difference-In-Difference," "Event Study," and "Quantile Treatment Effects" methods in a reduced-form analysis to compare outcomes between pre-reform linked physician peers and pre-reform unlinked physician peers. Overall, the results indicate that, prior to CRDS, there was a higher growth rate in patient referrals between linked physician peers compared to unlinked peers within the treatment group. However, following the implementation of CRDS, we observed an immediate decline in referrals between linked physician peers, alongside a slight increase in referrals between unlinked peers. This shift can be attributed to the centralization of the referral system introduced by CRDS, which allowed specialists to allocate more time slots to patients referred through the system, thereby reducing the reliance on direct referrals from general practitioners. In conclusion, the CRDS implementation appears to have facilitated a more equitable distribution of patients among specialist physicians.

• Peer effect in adolescent's mental health: the role of peer support:

Forthcoming Working Paper

This paper investigates the role of peer support in adolescents' depression, accounting for unobserved heterogeneity among adolescents. I use a structural model for the joint probability of the adolescents' depression score and the emotional support they seek from friends. The parameters of the model are estimated using a Bayesian approach with a data augmentation technique. I use the National Longitudinal Study of Adolescent Health in Adulthood (Add Health) data on adolescents in grades 7-12 during the 1994-95 school year in the US. I find that depression leads adolescents to seek more enacted support from friends and approximately 30% of adolescents have a latent demand for peer support greater than their total number of friends. I find that including the unobserved heterogeneity does not affect the effect of adolescents' depression on peer support in general. However, when estimating gender heterogeneity, do not include unobserved heterogeneity among females overestimates the effect of depression on peer support seeking. I note that females have a greater latent demand for support than males. Adolescents seek support from friends who are likely female, have a higher grade at school, and have higher raw picture vocabulary test scores. Using a single linear-in-mean model to estimate the endogenous peer effect in peer support, I find that adolescents with more friends who are willing to ask for support when depressed have a greater number of friends with whom they ask for support when they are depressed.

• The Effect of Incentive and Coercive Policies on Physician Retention in an Isolate or a Remote Region of Quebec:

With Bernard Fortin

Forthcoming Working Paper

This paper analyzes the effect of incentive and coercive measures on the retention of general practitioners in isolated and remote regions of Quebec. We use administrative records detailing the geographical distribution of general practitioners within Quebec from 1975 to 2021. This dataset is sourced from the Canadian Institute for Health Information (CIHI). We model the transition hazard rates from one region to another using a multiple spell multiple-states duration model with a mixed proportional hazard structure (MPH). The results of the model, which account for unobserved heterogeneity, suggest that different remuneration and bursary programs significantly decrease the risk of physicians transitioning from isolated or remote regions to other areas. These effects are further reinforced by coercive measures. Our findings indicate that including physicians' unobserved heterogeneity is crucial, and removing the different compensation programs could reduce the expected duration that a general practitioner remains in isolated and remote regions by about nine years.