Developing a theory of decision making in the face of unknown unknowns

By Marie-Louise Viero, Queen’s University

“There are known knowns: There are things we know we know. We also know there are known unknowns. That is to say, we know there are some things we do not know. But there are also unknown unknowns. The ones we don’t know we don’t know.”

–Donald Rumsfeld, former U.S. Secretary of Defense

Until recently, economic analysis has only been able to deal with known knowns and known unknowns. According to the way uncertainty has been modeled, economic agents (e.g. consumers, managers, investors, etc.) start out with a detailed understanding of all things that can possibly happen. When the agent receives new information, it narrows down that list of possibilities, so that the agent’s knowledge becomes more precise over time. But, the set of possibilities never changes. Economic agents never discover new possibilities that they didn’t recognize before. There are no “unknown unknowns.”

Existing economic theory says nothing about how to deal with unknown unknowns. However, in the real world, unexpected things do happen–things that were at one point inconceivable.  Becoming accustomed to possibilities that were once inconceivable is part of both history and our own life experience. As Rumsfeld’s quote suggests, agents often recognize the possibility of unknown unknowns when making decisions. This is true in national defense (e.g. new strategies of terrorists) and industrial organization (e.g. competitors inventing new products).

Both at the personal and societal level, our universe of possibilities frequently expands as we become aware of new opportunities, and economic models should be able to capture this.

My recent research with Edi Karni from Johns Hopkins University incorporates unknown unknowns into economic models. In our framework, agents can make discoveries over time that expand their awareness.

In going from the abstract idea of unawareness and growing awareness to an operational economic framework there have been two main challenges: The first is with the framework itself, since it has been proven that the traditional modeling framework precludes unawareness. The second is to connect preferences, that is, the utility function, across different levels of awareness.

In our first article, titled ““Reverse Bayesianism”: A Choice-Based Theory of Growing Awareness,” which was published in the American Economic Review, we provide a framework for modeling the expanding universe of agents in the wake of growing awareness, and link preferences under different levels of awareness. We also develop utility functions that capture agents’ behavior as well as rules for updating agents’ beliefs as awareness grows. The updating is dubbed “Reverse Bayesianism” because it is the inverse of the Bayesian updating that is used in the traditional models.

In the first article, agents are assumed, at any point, to act as if they are completely aware of all aspects of the universe. However, if an agent in the past has discovered new aspects of the world, it is reasonable for her to believe that it may also occur in the future.

Our latest paper expand our framework to allow for this possibility and investigates the behavioral implications of the theory. One of the results is that an agent’s utility function now includes an attitude towards the unknown unknowns; hence we can talk about the agent showing fear or excitement about the unknown, just as we can talk about her being risk averse or risk loving.

Our results are widely applicable. If, for example, an investor shows fear towards the unknown, investments that are less likely to generate unforeseeable results are preferred over similar investments that are more likely to generate unforeseeable results. This can explain a number of observations in financial markets.

This article was adapted from an article that was published in the Queen’s Economics Department annual newsletter.