Collaborative Research:Dynamic Rational Inattention: Theory and Applications

Models of rational inattention have proved to be useful in understanding many economic phenomena, such as price and investment stickiness, asymmetric response of consumption to income shocks, and coordination failure. The proposed research will provide a new tool to analyze and interpret these - as well as other issues that pertain to economic agents' limitations on the ability to process information - via a new class of dynamic rational inattention models that have a recursive form. The key component in these models is an Infinite Horizon Attention Constraint (IHAC), which captures the intertemporal trade-off in acquiring information. An IHAC specifies what type of information is affordable today, and how the choice of what to learn today affects the possible choices of information tomorrow and in the future. The IHAC is universal in a well-defined sense and yet it - as well as all the other parameters of the model, such as state dependent utilities and the anticipated evolution of beliefs over states - can be uniquely identified from choice behavior. First, the proposed research will characterize axiomatically the types of behavior that can be described by the model. Second, it will develop applications of the model that generate new theoretical predictions, both in the realm of individual decision making and in the context of strategic interactions, and in different areas, ranging from macro-economic phenomena to the explanation of evidence collected by psychologists and economists in laboratory experiments. For example, the model can capture notions such as expertise and fatigue in acquiring information and gives rise to new behavioral phenomena such as familiarity bias. It will also generate new insights about policies that might ameliorate inefficient attention choices. The relevance of this issue is underscored by the current debate about the role exhausted attention seems to play in trapping people in poverty (see Mani et. al. 2013 and Mullainathan and Shafir, 2013).