A new paper by Beshears, Choi, Laibson, Madrian and Milkman:
We measure how receiving information about coworkers’ savings behavior affects recipients’ savings choices. Employees of a large company who were not participating in or contributing little to the company’s retirement savings plan were sent a simplified enrollment or contribution rate increase form. A randomized subset of forms included information on the fraction of coworkers either participating in or contributing at least 6% of pay to the plan. We find that peer information increased savings of non-unionized recipients but decreased savings of unionized recipients. Our results highlight the possibilities and limitations of peer information interventions
The interaction with unions is perplexing. Non-union members obviously view others behavior as a positive cue (herding). Unionized recipients almost seem to see it as a opportunity for free-riding.
I think this is more of a “devil is in the details” result than a true limitation. You always have to be aware of non-nudge factors when nudging.