This article models the riskiness of structured securitization deals. The deals are put together by "banks," which can exercise strategic options over the risk put into the deals. The banks face a trade-off between the benefits of risktaking now and future franchise benefits if the deal pays off. The key insight is a convex relationship between the value of the bank's equity position and the risk in the deal. Although there is a continuum of possible risk, banks choose either the highest or lowest levels of risk open to them. Changes in strategy are discontinuous and unpredictable; a history of low risk-taking may be a prelude to increased risk-taking later. Competition, to the extent of reducing franchise value, can lead to more risk-taking, as can more information in the market. The model provides insights into the risk-taking that led up to the Great Recession and to institutions that are "Too Big to Fail." Risk-taking, particularly abrupt changes in risk-taking, was at the core of the recent boom and bust in U.S. real estate markets. This was manifested in the market for mortgage-backed securities, where there was a rapid change in the structure of the mortgage market from 2003 to 2005, as the growth of lower quality "private label" securities accelerated and there was an increase in mortgage types that made risk-taking easier (e.g., loans without full documentation; see Anderson, Capozza andVan Order 2011 andAshcraft andSchuermann 2008).Financial institutions have the ability to exercise strategic options when the holders of their liabilities have limited information and limited control over risk-taking. We focus on two options: (1) choice of assets so that the volatility of the portfolio backing the liabilities is higher than what the market believes and (2) the option to change strategy in response to changes in data. Our focus here will be on structured securitization, but similar analysis can be made of the behavior of financial institutions like banks and government-sponsored enterprises (GSEs) like We develop a model of risk-taking in securitization deals in which deal sponsors optimize the value of their options when putting together the deals. The model has a "bang-bang" solution; the structure either takes on a high level of risk or low level of risk even though there are feasible levels in between. This is not new in corporate finance models. For instance, Ross (1997) and Leland (1998) develop models of corporate risk-taking that have similar properties. In their models, the solutions depend on taxes and bankruptcy costs. Firms that would ordinarily take as much risk as they could get away with in order to maximize the shareholder value take less risk because bankruptcy costs increase the cost of risk-taking. This leads to a critical level of bankruptcy cost above which firms choose the safe strategy and below which they choose the risky strategy, which leads to "bang-bang" solutions as cost moves above or below the critical level. However, those models are not applicable to structured deals in whi...