2016 INFORMS Annual Meeting Program

TC18

INFORMS Nashville – 2016

TC18 106A-MCC

investment decisions, reduces his risk exposure as he becomes more sensitive to risk, and that his strategy depends non-monotonically on the aggregate risk level. 2 - Resource Allocation For Hepatitis C Elimination

Finance, Portfolio Contributed Session

Qiushi Chen, Georgia Institute of Technology, Atlanta, GA, United States, chenqiushi0812@gatech.edu, Turgay Ayer, Jagpreet Chhatwal

Chair: Christopher M Rump, Associate Professor, Bowling Green State University, College of Business Administration, Bowling Green, OH, 43403-0267, United States, cmrump@bgsu.edu 1 - A Goal Programming Approach To Municipal Bond Portfolio Management Laura Ventura, PhD Student, The Pennsylvania State University, University Park, PA, 16802, United States, ljv115@psu.edu, Barbara Venegas Quintrileo The consequence of the municipal bond tax-exemption is that retail investors represent the overwhelming majority of municipal bondholders. Retail investors’ buy and hold strategy results in low portfolio turnover causing limited inventory levels and obscure historical pricing that render modern portfolio theory unsuitable. In exchange we propose a non-preemptive goal programming model for municipal bond portfolio management. We consider a municipal bond index replication strategy using Morningstar’s municipal bond index data and Bloomberg’s municipal bond market data. The model determines bond selection that meets risk and return metrics sensitized to the number of transactions. 2 - Asset Selection In Indian Stock Market Using PCA-DEA Framework Dhanya Jothimani, Doctoral Student, Indian Institute of Technology Delhi, DMS, IIT Delhi, Vishwakarma Bhawan, Hauz Khas, New Delhi, 110016, India, dhanyajothimani@gmail.com, Ravi Shankar, Surendra S Yadav Portfolio optimization has three important stages. Among them, asset selection is the first and important stage. We use a Principal Component Analysis - Data Envelopment Analysis (PCA-DEA) framework for asset selection in Indian stock market. The sample consisted of firms listed in National Stock Exchange. The contributions are two-fold: first, the framework helps to avoid the curse of dimensionality of DEA and second, it aids in selection of asset for the second stage of portfolio optimization. 3 - Optimal Portfolio Under Black Litterman Framework With Certain Confidence Level Cagatay Karan, North Carolina State University, Raleigh, NC, United States, ckaran@ncsu.edu, Tao Pang Under the Black-Litterman framework, the investor’s views can be integrated with the classical mean-variance portfolio optimization in a Bayesian manner. Typically, the investor is not 100% sure about her view, so the confidence level of the view plays an important role in determining the optimal portfolio. We propose a simple but meaningful method based on the investor’s confidence level on whether the market is a bull market. Conditional Value at Risk (CVaR) is used as the risk measure instead of variance, and mixed Gaussian distributions are used to model the assets’ market returns. The optimal portfolio is explicitly obtained from the optimal portfolio weights under the proposed setting. 4 - Evolution Of A Lottery Jackpot Christopher M Rump, Associate Professor, Bowling Green State University, College of Business Administration, Bowling Green, OH, 43403-0267, United States, cmrump@bgsu.edu We develop a predictive model for the growth of the jackpot prize in large, multi- state lotteries. The prediction is based on ticket sales inferred from the number of lesser prizes awarded after each lottery drawing. With this jackpot growth model, we investigate whether or not this gamble ever has positive expected value and make recommendations for the best time to play the lottery if you must. TC19 106B-MCC Population Health: Infectious and Chronic Diseases Sponsored: Computing Sponsored Session Chair: Nedialko Dimitrov, The University of Texas at Austin, The University of Texas at Austin, Austin, TX, 00000, United States, ned.dimitrov@gmail.com 1 - Risk Sensitive Control And Cascading Defaults Agostino Capponi, Columbia University, ac3827@columbia.edu We consider an optimal risk-sensitive portfolio allocation problem, which explicitly accounts for the interaction between market and credit risk. The investor allocates his wealth on a portfolio of assets, which can default sequentially and cause distress to the remaining assets in the portfolio. We give an explicit characterization of the optimal feedback strategies. A numerical analysis indicates that the investor accounts for contagion effects when making

More than 170 million people are infected with hepatitis C virus (HCV) globally. The recent availability of highly effective treatments offers an opportunity to control current epidemic and eliminate HCV worldwide. However, high drug cost and unawareness of infection are challenges for achieving this goal. In this study, we develop an HCV transmission model, and identify optimal allocation of resources towards HCV screening and treatment to achieve the disease control target at the minimum cost. We present the allocation policies in different health care settings and target population profiles. 3 - Optimizing Arbovirus Surveillance Xi Chen, University of Texas at Austin, carol.chen@utexas.edu We introduce a county-level risk assessment framework for identifying areas that may be at high risk for importation of arboviruses. Human importation risk is estimated using a maximum entropy algorithm, based on historical dengue importation data, socioeconomic, demographic, and bio-climatic data. A significant reason for the popularity of the maximum entropy methodology is its applicability to presence-only data. To address the uncertainty quantification in the point estimation of maximum entropy model, we analytically derive an expression of the variance of the target species distribution probabilities and comparing the results with bootstrap methods.

TC20 106C-MCC Multiagent Systems Modeling

Invited: Tutorial Invited Session

Chair: Sanmay Das, Washington University in St. Louis, St, Louis, MS, 12, United States, sanmay@wustl.edu 1 - MultiagentSystems Modeling Sanmay Das, Washington University in St. Louis, St, Louis, MS, United States, sanmay@wustl.edu

A multiagent system is one where multiple autonomous agents with potentially different goals interact. Viewing agents through the computational lens provides a powerful, yet principled method for understanding the behaviors of complex systems, including economic and financial markets, online social networks, etc. In this tutorial, I discuss general principles for such modeling, best practices for handling the simplicity/complexity tradeoff, and present examples of predictive and useful models. TC21 107A-MCC Payment Models, Pricing, and Incentives in Healthcare Sponsored: Health Applications Sponsored Session Chair: Mehmet U.s. Ayvaci, University of Texas at Dallas, Richardson, TX, United States, mehmet.ayvaci@utdallas.edu 1 - The Role Of Physician Alignment And Organizational Structures In Bundled Payments Jan Vlachy, Georgia Institute of Technology, Atlanta, Georgia, vlachy@gatech.edu, Turgay Ayer, Mehmet U.S. Ayvaci, Srinivasan Raghunathan Bundled payments in healthcare unify the payments to care providers. Motivated by the low rates of voluntary bundling, we formulate game-theoretic models to understand the incentives of hospitals and physicians when forming a bundle. Our analyses lead to several interesting findings with policy implications: 1) alignment between the hospital management and physicians is critical in successful bundling, 2) integrated hospital systems or hospitals with salaried physicians are likely to benefit more from bundling, and 3) under the current bundled payment mechanism, overall care quality may decrease. We further propose alternative designs to ensure sufficient quality.

308