Incentivizing Compliance with Algorithmic Instruments

           Daniel Ngo * 1 Logan Stapleton * 1 Vasilis Syrgkanis 2 Zhiwei Steven Wu 3

             Abstract                 causal effects rely on randomized experiments, which ran-
   Randomized experiments can be susceptible to         domly assign each individual in a population to a treatment
   selection bias due to potential non-compliance by      group or a control group and then estimate the causal effects
   the participants. While  ...