Abstract
We extend Garicano's (2000) model of optimal organizations by allowing its members to screen problems, i.e. to attempt the identification of problems before trying to solve them. As for solving problems, screening is costly to learn and time consuming but has the advantage of allowing for successfully screened problems to be directed to those in the organization who can solve them. We establish several properties of optimal organizations and use them to show: (a) When screening problems is as costly as solving them, optimal organizations are hierarchies as in Garicano (2000), but (b) when the cost of learning how to screen problems is sufficiently small, optimal organizations are such that workers screen all problems that they and the managers who solve the most extraordinary problems cannot solve, those problems that they screen are directed to those managers who can solve them and those problems that they neither solve nor screen are passed to the managers who solve the most extraordinary problems. For intermediate values of the cost of learning how to screen problems, we show computationally that the optimal organization is a hybrid of the above two organizational forms. * We wish to thank Paulo Bastos, Esteban Rossi-Hansberg, John Van Reenen, three anonymous referees and seminar participants at the 2021 RES annual conference (Belfast) and EWET 2021 (Akko) for helpful comments. Any remaining errors are, of course, ours.