Switching Forth and Back: Problem Solving under Time Pressure
48 Pages Posted: 6 Dec 2019
Date Written: November 20, 2019
We study optimal problem solving under deadlines. An agent dynamically allocates effort between a trial-and-error approach and a structural approach. The trial-and-error approach requires only a single step but may be infeasible. The structural approach is always feasible but requires more steps. The unique optimal policy can have the switching-forth-and-back property: the agent switches back to trial and error after she left it for the structural approach. We identify three drivers of that optimal policy: costly effort, finite deadlines, and observability of intermediate progress. Our model applies to questions in organizational economics, innovation, and education.
Keywords: Dynamic Problem Solving, Exponential Bandits, Deadlines
JEL Classification: D01, D83, O31
Suggested Citation: Suggested Citation