Abstract I analyze the time consistency of ex-ante optimal strategies with bounded memory, in a general class of stationary dynamic environments called Markov Decision Processes with Partial Observation (POMDP), as in Kalai & Solan (2003). In games with absent-mindedness an ex-ante optimal strategy may not be time consistent. However, Piccione & Rubin- stein (1997) shows that it satisfies a weaker form of time consistency, i.e. modified multiself consistency, where the agent would not deviate from the strategy for the cur- rent period assuming he will follow it from tomorrow on. I show that in any such environment an ex-ante optimal strategy with bounded memory is time consistent in this weak sense, hence no tension for ex-ante versus interim incentives arise in carrying out the optimal bounded memory strategy.“
Short Bio :
Yilmaz Kocer received his Ph.D. in Economics from NYU in 2010. He did his postdoctoral study at Princeton in the Economic Theory Center, between 2010-2011 working on Incentived Learning in a Principal-Agent framework. He then joined USC as an Assistant Professor of Economics in the Department of Economics in 2011. His dissertation is on a theory of single agent learning under memory constraints. His broader research interests are Bounded Rationality and Decision Making, Incentives and Learning, Behavioral Economics and Game Theory.