ABSTRACT
The search for clean identification strategies in empirical economics has increased concerns for a lack of external validity. In the presence of only true feedback, this would not be a problem, as it would provide us with rich information about economic reality. This information would in turn allow us to reliably falsify and improve our economic theories. More local estimates may also be more favorable for policy makers, as they provide them with information about their immediate environments. In the presence of false feedback, however, we run into an identification problem, as we are not able, anymore, to separate variation due to the context of the study from variation due to the garden of forking paths. The chapter thus argues that we need to fix the latter problem before we can tackle the former.
