A manager wants to know if employees who are given relative performance evaluations work longer hours than employees who are not evaluated relative to their colleagues.
a) Explain how an experiment could be used to estimate the causal effect of relative performance evaluations on hours of work.
b) Instead of running an experiment, the manager decides to estimate the effect of relative performance evaluations on hours by comparing average hours of work across firms that use relative performance evaluations and firms that do not use relative performance evaluations.
Under what conditions will this comparison tell the manager about the causal effect of relative performance evaluations on hours of work?