I would like to see this replicated, but the result is interesting nonetheless:
Using US panel data on young workers, we demonstrate that those who receive performance pay are more likely to consume alcohol and illicit drugs. Recognizing that this likely reflects worker sorting, we first control for risk, ability, and personality proxies. We further mitigate sorting concerns by introducing worker fixed effects, worker-employer match fixed effects, and worker-employer-occupation match fixed effects. Finally, we present fixed effect IV estimates. All of these estimates continue to indicate a greater likelihood of substance use when a worker receives performance pay. The results support conjectures that stress and effort increase with performance pay and that alcohol and drug use is a coping mechanism for workers.
By Benjamin Arta, Colin P. Green, and John S. Heywood, via the excellent Kevin Lewis.