UC Berkeley

This dissertation investigates the relationship between the use of financial risk taking and the complex mathematical models used to quantify risks. In the first essay, laboratory experiments demonstrate that students who have taken courses in finance and economics take more risk when they are exposed to complex mathematical models, even when these models do not provide more information. Furthermore, students who have a strong belief in the power of risk quantification to produce accurate assessments of future events are more likely to take more risk when it is accompanies by complex quantitative models. The second essay uses data on U.S. banks from 1994-2007 and investigates popular claims that innovative risk measures, such as value at risk (VAR) resulted in greater risk seeking by banks prior to the financial crisis. It theorizes that as formal risk assessment models, such as the risk models used by banks, become institutionalized within organizations, a model's abstract representation of reality becomes reified and treated as though it is real and complete. This results in organizationally-embedded decision-making heuristics that shape how choices are made within the firm. For risk models, this meant that when banks encouraged the use of risk models, they generated an implicit belief that these models represented accurate assessments of future outcomes. This reduced uncertainty about future outcomes and led to greater risk taking.