A low-key social insurance reform - treatment effects for back pain patients in Norway

This paper estimates treatment effects for back pain patients using observational data from a low-key social insurance reform in Norway. Using a latent variable model we estimate the average treatment effects (ATE), the average effect of treatment on the treated (TT), and the distribution of treatment effects for outpatient treatment at three different locations. To estimate these parameters and the distribution of treatment effects we use a discrete choice model with unobservables generated by a factor structure model. Distance to nearest hospital (in kilometers) is used as an instrument in estimating the different treatment effects. We find a positive effect of treatment of 6 percentage points on the probability of leaving sickness benefits after allowing for selection effects and full heterogeneity in treatment effects. We also find that there are sound arguments for increasing the outpatient program of treating back pain patients.