The Decision Making of Risk and Uncertainty

Abstract

In the last half century or so, the expected utility model has become the work horse model that economists fall back upon when modeling the choices made by decision makers faced with uncertainty in the environment. Examples of its application range from the ‘dating market’ to the financial market. My five lectures will be an introduction to the expected utility model—its foundations, as well as, some of its limitations. The origin of the expected utility model goes back to the 1730s. Daniel Bernoulli proposed this model as a way of accommodating the famous St. Petersburg Paradox. The modern treatment of the model starts with the pioneering work of vonNeumann and Morgenstern. They provided an axiomatic foundation for this model. Their treatment assumed a situation of risk, that is, a situation in which the uncertainty faced by a decision maker can be captured via objective probabilities. For example, consider a decision maker who is faced with a gamble that gives her $100 if an unbiased coin toss shows up heads and $0 if it shows up tails. Here the decision maker may summarize the uncertain prospect that she is faced with using objective probabilities, namely, she receives $100 with probability 0.5 and $0 with complementary probability. As opposed to this we can imagine a situation of ‘pure uncertainty.’ Consider changing the gamble that the decision maker faces to the following one: She will get $100 if a South American team wins the next soccer world cup and nothing otherwise. Is such a decision maker able to attach a unique probability to an event like a South American team winning the next world? And, if so, are the choices that Originally published in 1738 in the Commentaries of the Imperial Academy of Science of Saint Petersburg ; translated by Sommer, Lousie (1954): “Exposition of a New Theory on the Measurement of Risk,” Econometrica 22 (1): 22 – 36.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 0 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.