Quantifying Causal Effects for Continuous Treatments via a Mixed Model Generalized Propensity Score Estimator

• Activity Number: 607
• Type: Contributed
• Date/Time: Thursday, August 7, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Business and Economic Statistics Section
• Author(s): Daniel Graham*+ and David Stephens and Emma McCoy
• Companies: Imperial College London and McGill University and Imperial College London
• Keywords: Causal ; propensity score ; longitudinal ; transportation

Abstract

This paper develops a mixed model generalised propensity score (PS) estimator to quantify the causal effect of continuous treatments. The motivation is that we seek to estimate a dose-response relationship but suspect confounding from both observed and unobserved characteristics. Analytical results and simulations show that by specifying unit level random effects within a longitudinal generalized PS model we can adjust effectively for time-invariant unobserved confounding, but more extensive conditioning can adversely affect efficiency and can render the task of finding overlap in support of the covariate distribution more challenging. We apply the estimator to study the effects of road network capacity expansions on aggregate urban traffic volume and density in US cities. Our results show that network capacity expansions do cause substantial increases in aggregate urban traffic volumes. The induced travel demand effect is sufficiently strong that traffic densities continue to grow over a vast range of capacity increase and congestion remains unaffected. These results have important implications for optimal urban transportation strategies.