Informs Annual Meeting 2017

MC55

INFORMS Houston – 2017

2 - A Decision Making Approach for Crop Seed Variety Selection via Hedging Against Weather Risk Leyuan Shi, Peking University, Room 512 Founder Building, 298 Chengfu Road, Beijing, 10080, China, leyuan.shi@wisc.edu Leyuan Shi, University of Wisconsin, Madison, WI, United States, leyuan.shi@wisc.edu, Zhongshun Shi, Yu Zhao, Xi Zhang This paper develops an efficient decision making approach for the seed retailer to determine the seed varieties selection in appropriate proportions for the next season. The proposed approach integrates the customized portfolio selection theory and effective machine learning technique. We formulate this decision making problem as the mean-covariance model to balancing the weather risk and crop yield. A hybrid coefficients evaluation method is designed to produce the expected yield of each seed variety under different weather type. We develop a MIQP-based decomposition algorithm to solve the decision making problem efficiently and effectively. 3 - A Machine Learning Approach to the Syngenta Crop Challenge 2017 Wenjun Zhou, University of Tennessee, Knoxville, TN, 37996, United States, wzhou4@utk.edu, Yunhe Feng Seed companies need to plan for the variety and quantity of seeds to stock at least a year in advance. There are a large number of seed varieties, each can perform best under different growing conditions. Since next year’s weather is unpredictable, it is hard to foresee which seed varieties work the best in each field. We propose using multi-task learning for estimating the yield and risk of each variety as if they were planted at each location. The best mix of seeds for each location was then determined by seeking a trade-off between yield and risk. 362B Topics in Finance: Systemic Risk Management Sponsored: Financial Services Sponsored Session Chair: Xuedong He, xdhe@se.cuhk.edu.hk Co-Chair: Nan Chen, nchen@se.cuhk.edu.hk 1 - Submodular Risk Allocation Paul Glasserman, Columbia University, 403 Uris Hall, Columbia Business School, New York, NY, 10027, United States, pg20@columbia.edu, Samim Ghamami Changes in the over-the-counter derivatives markets leave dealers some flexibility in whether to trade bilaterally or through central counterparties (CCP) and which CCP to choose. We analyze the problem of optimal allocation of trades to portfolios to minimize risk-based margin and capital requirements. With submodular margin requirements, the problem becomes a submodular intersection problem; the dual provides per-trade margin attributions. We derive conditions under which standard deviation and other risk measures are submodular functions of sets of trades. We compare systemwide optimality with individually optimal allocations. 2 - On Default Probabilities in Financial Networks We consider the Eisenberg-Noe network model for systemic risk, focusing on random shocks to financial institutions. Using duality, we characterize shock amplification caused by the network structure and find when a specific bank fails. These results enable us to improve our understanding of default probabilities in financial networks. Specifically, we establish general upper and lower bounds of firm-specific default probabilities. We also discuss the impact of network structure on default probabilities. In addition, as default events become rare, asymptotic behaviors of corresponding probabilities can be derived. 3 - Financial Contagion in a Multilayer Network Zachary Feinstein, Washington University-St. Louis, St. Louis, MO, United States, zfeinstein@wustl.edu This talk provides a general framework for modeling financial contagion in a system with obligations in multiple illiquid assets. In so doing, we develop a multi-layered financial network that extends the single network of Eisenberg and Noe (2001). In particular, we develop a financial contagion model with fire sales that allows institutions to both buy and sell assets to cover their liabilities and act as utility maximizers. We prove that, under standard assumptions, equilibrium portfolio holdings and market prices exist which clear the multi-layered financial system. We demonstrate the value of our model through illustrative numerical case studies. MC55 Dohyun Ahn, PhD Candidate, KAIST, #2111, E2-2, KAIST, 291 Daehak-ro, Yuseong-gu, Daejeon, Korea, Republic of, dohyun.ahn@kaist.ac.kr, Nan Chen, Kyoung-Kuk Kim

4 - Leverage Constraint, Market Illiquidity, and Financial Fragility Nan Chen, Chinese University of Hong Kong, William M.W. Mong Engineering Bldg, Rm 609, Shatin N.T, Hong Kong, nchen@se.cuhk.edu.hk, Jing Chen, Xuedong He We establish a full equilibrium model to show how systemic fragility is built up in the presence of leverage constraints and market illiquidity. The model can generate two distinct regimes: normal and crisis. Due to highly nonlinear amplification effects caused by both illiquidity, the economy is prone to instability and occasionally enters into the volatile crisis regime. The equilibrium risky asset price demonstrates heteroscedasticity in different regimes. Our model shows that margin requirement, a common practice in risk management at the micro level, may yield an undesirable pro-cyclic effect in a systemic context. 362C Joint Session SOLA/PSOR: Public Sector Location Analysis Sponsored: Location Analysis Sponsored Session Chair: Kayse Lee Maass, Mayo Clinic, Rochester, MN, 7, United States, maass.kayse@mayo.edu 1 - A Robust Model for Pre-positioning Relief Items for a Typhoon Joline Uichanco, University of Michigan, Ross School of Business, 701 Tappan St, Room 4418, Ann Arbor, MI, 48109-1234, United States, jolineu@umich.edu We consider the problem faced by a humanitarian agency in pre-positioning relief items to prepare for an oncoming typhoon whose future outcome (trajectory and wind speed) is uncertain. We develop probabilistic models of municipality-level demand and of damage to pre-positioned supply that are dependent on the typhoon outcome. The dependence of demand on the typhoon outcome is determined by a hierarchical linear model estimated from a dataset that includes typhoon effects from five West Pacific typhoons affecting the Philippines. We propose to solve the pre-positioning problem using a bi-objective two-stage stochastic optimization model. 2 - Anticipation Effects in Post-disaster Supply Chain Modeling Diana Gineth Ramirez-Rios, Rensselaer Polytechnic Institute, 134 25th Street, Troy, NY, 12180, United States, ramird2@rpi.edu, Jose Holguin-Veras, Luk N. Van Wassenhove, Victor Cantillo, Shaligram Pokharel, Johanna Amaya-Leal, Trilce Encarnacion This research focuses on disaster relief modeling, particularly in the period of time starting at the moment a disaster strikes and when the initial response takes place. This research introduces the anticipation function, which assesses both the deprivation and psychological cost or anxiety of the deprived individual. The results of an exploratory analysis of this anticipation function is empirically derived through SP surveys. Some management implications and its possible incorporation to the supply chain model formulation were also considered. 3 - Bike Sharing using Toggle Locations and Portable Stations Rahul Swamy, University of Illinois at Urbana-Champaign, 205 E. Clark St, Apt 303, Champaign, IL, 61820, United States, rahulswamy91@gmail.com, Jose Luis Walteros We explore an alternative approach to fixed station locations in a bike sharing system by turning on/off (toggle) certain locations for certain periods of time. A Mixed Integer Program is presented that decides which of the candidate locations to choose to serve as bike stations at what time, with the objective of minimizing the bike redistribution load. The MIP has a rich structure that is exploited using Bender’s Decomposition. The second part of the talk focuses on how portable stations can be used to serve in the toggle station locations. This routing problem is solved using implicit enumeration followed by a variant of set covering problem. Computational results are presented with a Case Study in NYC. 4 - Selecting Locations of Rehabilitative Shelters for Human Trafficking Survivors among States in the U.S.: A Decision Analytic Approach Kayse Lee Maass, Mayo Clinic, Rochester, MN, United States, maass.kayse@mayo.edu, Renata Alexandra Konrad, Andrew C. Trapp Rehabilitative shelters are an important component of anti-human trafficking efforts and entail a large investment. While survivors of human trafficking exist in every U.S. state, roughly only 2/3 of states provide dedicated shelters. Even in states with shelters, demand exceeds the current capacity. Furthermore significant disparity exists between states with regard to the legislative environment and provision of auxiliary services for survivors. This project develops a decision analytic approach to evaluate the impact of a dedicated shelter at the state level. Using concepts from health and social welfare economics, we develop a budget allocation problem that maximizes societal value. MC56

222