2016 INFORMS Annual Meeting Program

SC43

INFORMS Nashville – 2016

SC43 208A-MCC Spatial Risk and Decision Analysis Sponsored: Decision Analysis Sponsored Session Chair: Gilberto Montibeller, Loughborough University, Loughborough, United Kingdom, g.montibeller@lboro.ac.uk 1 - Spatial Risk Analysis In Emergency Management Nikolaos Argyris, Loughborough University, Loughborough, United Kingdom, n.argyris@lboro.ac.uk, Simon French In any emergency, there is a great deal of uncertainty, often geographical or spatio-temporal uncertainty. For instance, an industrial accident may lead to a plume of contamination, putting populations at risk downwind. The path of a hurricane provides another example that is of obvious concern to emergency managers. We consider how analysts can communicate spatio-temporal uncertainty to those handling the crisis. We review the somewhat limited literature on the representation of spatial uncertainty on maps. We note that many cognitive issues arise and that the potential for confusion is high. We then make some suggestions based upon the idea of presenting multiple scenarios. 2 - Spatial Preference Functions For Risk Analysis Jay Simon, American University, jaysimon@american.edu, L Robin Keller When outcomes are defined over a geographic region, measures of spatial risk regarding these outcomes can be more complex than traditional measures of risk. One of the main challenges is the need for a cardinal preference function that incorporates the spatial nature of the outcomes. We explore preference conditions that will yield the existence of spatial measurable value and utility functions, and discuss their application to spatial risk analysis. 3 - Multi-criteria Spatial Risk Analysis For Resource Allocation Decisions Gilberto Montibeller, Full Professor of Management Science, Loughborough University, Loughborough University, United Kingdom, g.montibeller@lboro.ac.uk, Valentina Ferretti There is a broad literature on spatial multi-criteria evaluation in the environmental domain and some attempts of conducting risk analysis in this context. Most of these attempts neither employ a proper decision analytical framework nor provide a clear conceptualization for allocating resource on mitigating actions. To address these weaknesses we conceptualize a multi-criteria spatial risk analysis assessment, which may support spatial decision-making processes. The framework employs expected multi-attribute utility and portfolio decision analysis concepts in a spatial context. A case study on flooding evaluation and defense building illustrates its application in practice. SC44 208B-MCC Modeling of Uncertainty and Preference in Decision Analysis Sponsored: Decision Analysis Sponsored Session Chair: Christopher Hadlock, Austin, TX, United States, cchadlock@gmail.com Co-Chair: Robert Hammond, Chevron, Houston, TX, United States, robertkh@gmail.com 1 - Johnson Quantile-parameterized Distributions Christopher Hadlock, The University of Texas at Austin, cchadlock@gmail.com, J. Eric Bickel It is common practice in decision analysis to elicit quantiles of continuous uncertainties, and then fit a continuous probability distribution to the corresponding probability-quantile pairs. This process is inconvenient because it requires access to a curve-fitting process, and the best-fit distribution will often not honor the assessed points. By strategically extending the Johnson Distribution System, we design the new J-QPD distribution system, which is directly parameterized by and honors any symmetric percentile triplet of low-base-high assessments in conjunction with known support bounds, eliminating the need to apply a fit procedure. 2 - The Metalog Distributions Thomas Keelin, Keelin Reeds Partners, 770 Menlo Avenue, Suite 230, Menlo Park, CA, 94025, United States, tomk@keelinreeds.com The metalog distributions constitute a new system of continuous univariate probability distributions designed for flexibility, simplicity, and ease of use. The system includes quantile-parameterized unbounded, semi-bounded, and bounded distributions, each of which offers shape flexibility that compares favorably with

Pearson distributions and others. Applications in fish biology and hydrology show how metalogs enable unprecedented insight into CDF data. A decision analysis application shows metalogs aided a decision that would have been made wrongly based on traditional discrete methods. 3 - Reexamining The Viability Of Scoring Rules Zachary Smith, The University of Texas at Austin, zack.smith@utexas.edu, J. Eric Bickel There are a number of widely used proper scoring rules used to elicit and rank expert opinions. However, not all rules have the property of being additive, in the sense that the score for marginal distributions and joint distributions are comparable. Scoring rules without this property are sensitive to the presentation of information as well as the information itself. We characterize scoring rules that are additive, and consider practical implications for some commonly-used rules. SC45 209A-MCC Panel: Systemic Risk Issues in Counterparty Risk and Central Clearing Invited: Risk and Compliance Invited Session Moderator: Agostino Capponi, Columbia University, 500 West 120th street, New York, NY, 10027, United States, ac3827@columbia.edu 1 - Panel on Systemic Risk Issues In Counterparty Risk And Central Clearing Agostino Capponi, Columbia University, ac3827@columbia.edu The panel is formed by six leading experts in the area of systemic risk and central clearing counterparties. The discussion will be centered on the economics of clearinghouses and their role in promoting financial stability. Pro and cons of central clearing will be highlighted and possible unintended consequences will be discussed. 2 - Panelist John Birge, Chicago Booth School of Business, jbirge@chicagobooth.edu 3 - Panelist Akhtarur Siddique, Office of Comptroller of Currency, Akhtarur.Siddique@occ.treas.gov SC46 209B-MCC Dynamic Pricing with Substitution, Learning and Reference Price Effects Sponsored: Revenue Management & Pricing Sponsored Session Chair: Candace Arai Yano, University of California-Berkeley, IEOR Dept. and Haas School of Business, Berkeley, CA, 94720, United States, yano@ieor.berkeley.edu 1 - Optimal Use And Replenishment Of Substitutable Raw Materials In Non-Stationary Capacitated Systems With Dynamic Price Izak Duenyas, University of Michigan-Ann Arbor, duenyas@umich.edu We consider a make-to-order setting where a firm can use either of two kinds of materials (or their mixture) to produce an end product using a shared production line with stochastic capacity. The materials are substitutable but one has a higher conversion rate and the other is cheaper, and their availability is uncertain. We show that a Use-down-to/Balancing Production Policy and modified Order-up-to Ordering Policy is optimal. Although the optimal policy is hard to compute using brute-force due to curse of dimensionality, we use its structure to develop an algorithm that solves for it efficiently. We also conduct sensitivity analysis of the optimal policy and find counter intuitive results. 2 - A Squared-coefficient-of-variation Rule For Learning And Earning N. Bora Keskin, Duke University, Durham, NC, United States, bora.keskin@duke.edu Consider a price-setting firm that sells products over a continuous time horizon. The firm is uncertain about the sensitivity of demand to price changes and updates its prior belief on an unobservable sensitivity parameter by observing demand responses. We derive and solve a PDE to show how the value of learning should be projected onto prices in an optimal fashion.

84