DP16911 Training a Sluggish System

Author(s): Kfir Eliaz, Ran Spiegler
Publication Date: January 2022
Keyword(s):
JEL(s):
Programme Areas: Organizational Economics
Link to this Page: cepr.org/active/publications/discussion_papers/dp.php?dpno=16911

Many organizational and biological systems need to maintain preparedness for external challenges. However, such systems tend to change their capabilities only gradually. How should we design training plans to enhance such systems' long-run preparedness? We present a model of optimal training plans for a rational, slowly adjusting system. A "trainer" commits to a Markov process governing the evolution of training intensity. At every time period, the system adjusts its "capability", which can only change by one unit at a time. The trainer maximizes long-run capability, subject to an upper bound on average training intensity. We consider two models of the system's adjustment: myopic/mechanistic and forward-looking. We characterize the optimal training plan in both cases and show how stochastic, time-varying intensity (resembling "periodization" techniques familiar from exercise physiology) dramatically increases long-run capability.