Top Highlights

• causal effects generally cannot be evaluated for a specific individual, an issue that Holland (1986) calls the “fundamental problem of causal inference.”
• the assumption that someone’s potential outcomes are not affected by the treatment status of others. This is known as the stable unit treatment value assumption (SUTVA); see, for instance, the discussion in Rubin (1980) and Cox (1958).
• However, it is important to acknowledge that ruling out any spillover or interference effects from the treatment of others to a subject’s own outcome might not appear plausible in all contexts.
• For instance, if more individuals obtain training, then the supply or availability of a certain skill in the labor market increases. This may negatively affect the wage of an individual independent of her or his own training participation
• if the number of trained individuals is small relative to the total demand for and supply of individuals with that skill, then the SUTVA might at least come close to being satisfied.