This approach strategically avoids
individual level selection bias but limits generalizability and introduces firm level selection bias.
In contrast, we take a previously unexplored middle road and compare the cost trends for all
individuals at firms offering CDHPs to cost trends for individuals at similar firms not offering
CDHPs. This is an intent-to-treat design as the focus is on individuals offered CDHPs and not on
individuals enrolled in CDHPs. Since the decision to offer CDHPs is made by the employer, this
intent-to-treat design eliminates any individual level selection bias. The design also likely
reduces any firm level selection bias as offering a CDHP alongside other plans is a moderate step
relative to full replacement with a CDHP