Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

MC41

4 - Designing Resilient Power System under Natural Hazards Tomas Ignacio Lagos, Universidad de Chile, 2017 - Pozuelo, Vitacura, Santiago, 7640031, Chile We propose an assessment framework to measure the energy-not-supplied of the Chilean electricity grid conditional to high-impact-low-probability hazards events, such as earthquakes. Then, we use an Optimization via Simulation approach for designing the best investment set on the system in terms of resilience. The simulator contains historical earthquake data, fragility curves of the network components, and a unit commitment model. We compare the results with traditional reliability-based methods and prove that these do not provide resiliency to the system. Finally, we show approximation methods that allow to improve the robustness to deliver optimal solutions using the heuristic. n MC40 North Bldg 226B Decision Making Under Uncertainty Sponsored: Decision Analysis Sponsored Session Chair: Youngsoo Kim, PhD, University of Alabama, Tuscaloosa, AL, United States 1 - An Interactive Bayesian Method for Multicriteria Sorting Problems We present a method that interactively places alternatives into preference categories (e.g. most, somewhat and least preferred). We assume the decision maker has a linear value function and her responses are prone to error. In each step of the algorithm, we ask the decision maker to place an alternative in a category and using this information, we update our estimate of her value function via MCMC. We compare our method to other algorithms in the literature that assume away response errors. 2 - The When and How of Delegated Search Sasa Zorc, INSEAD, 1 Ayer Rajah Avenue, PhD Office, Singapore, 138676, Singapore, Ilia Tsetlin, Sameer Hasija We consider the decisions of whether and how to outsource a search process (e.g., headhunting, search for investment opportunities). We model it in the dynamic principal-agent framework. The optimal contract is shown to have a purchase rights structure: the agent is paid a retainer plus a bonus if the principal accepts the search outcome. The size of this bonus can be defined a priori and is decreasing in time. The decision of whether to outsource or no can be reduced to the principal’s preferences over the quality-speed tradeoff, with quality preference leading to optimality of in-house search. 3 - Dual Reoptimization based Approximate Dynamic Programming Policies Danial Mohseni-Taheri, College of Business, University of Illinois at Chicago, 601 S. Morgan St., Chicago, IL, 60607, United States, Selvaprabu Nadarajah, Alessio Trivella Markov decisions processes (MDPs) with continuous endogenous and exogenous state and action spaces that are all high dimensional arise in operations applications. An example is the dynamic procurement of renewable power using multiple contracts under uncertain power prices and demand. The information relaxation and duality approach (IRD) can be used to obtain a lower bound on the optimal policy value of such MDPs, but approximate dynamic programming methods for computing operating policies are limited. We present a dual reoptimization method for extracting a feasible policy from IRD in this setting, guarantees on this policy, and numerical results on a renewable procurement application. 4 - Strategic Investment in Shared Suppliers with Quality Deterioration Youngsoo Kim, University of Alabama, Tuscaloosa, AL, United States, Dharma Kwon, Anupam Agrawal Firms often invest in their suppliers to improve their quality, but these suppliers are often shared by other firms who also consider investing in them. Motivated by this, we study a game of investment where one firm can free-ride on the other’s investment. Facing a continued deterioration in the supplier’s quality, each firm repeatedly decides when to invest in the supplier’s quality. We find that the repetitive nature of the investment induces inefficient delays in investment. We then compare the inefficient equilibrium to the first-best and estimate the resulting efficiency loss by using primary data, concluding that coordination among the firms can potentially save substantial amount of money. Canan Ulu, Georgetown University, McDonough School of Business, Operations and Information Management Area, Washington, DC, 20057, United States

5 - Protecting the Downside While Minding the Upside: Comparing Robustness Approaches in a Product Line Design Problem Tan Wang, University of Texas at Austin, Austin, TX, United States, Genaro J. Gutierrez Robustness approaches are useful when decisions must be made before uncertainties resolve. Two major approaches to robustness are: 1) maximizing the value of the worst outcome and 2) minimizing the ex-post regret of the decision made. We examine performance tradeoffs between the two approaches in the product line design problem, and we construct an efficient frontier for robust designs obtained by both approaches. Our findings indicate that it is possible to find robust designs that mitigate downside risk while limiting ex-post opportunity losses. n MC41 North Bldg 226C Healthcare Decision Analysis Sponsored: Decision Analysis Sponsored Session Chair: Mehmet Ayvaci, University of Texas at Dallas, Richardson, TX, United States Co-Chair: YeongIn Kim, University of Texas-Dallas, Richardson, TX, 75080, United States 1 - Examining Impacts of Clinical Practice Variation on Operational Performance Seokjun Youn, Texas A&M University, 320M Wehner Building, 4217 Texas A&M University, College Station, TX, 77840-4217, United States, Gregory R. Heim, Subodha Kumar, Chelliah Sriskandarajah Motivated by bundled payment policies that aim to reduce practice variation, this study examines whether and how lower variation in clinical practice relates to hospital operational performance. We further delve into the granular level of practice variation, such as the under-ordering risk of laboratory/radiology tests, to suggest actionable improvement plans. Using six-year inpatient data from NY and FL, we find that hospitals with higher practice variation tend to spend more resource for patient treatment. We also shed light on the intervening impacts of hospital quality evaluations on the relationship and deliver policy implications. 2 - Determinants of Health Information Exchange Use on Different Stages of Postadoption Xiang Wan, University of Florida, Gainesville, FL, United States, Emre M. Demirezen, Anuj Kumar The full potential of the benefits of health information exchanges (HIEs) cannot be realized until HIEs are used by physicians regularly. Few studies have focused on HIE postadoption phenomena and little is known about how different factors affect the continued HIE use. We empirically examine the effect of physician workload and disease complexity on HIE use at three different stages after physicians’ first use: namely initial use, moderate use, and heavy use. We find that the relationship between physician workload (or disease complexity) and continued HIE use behavior depends on the specific stage the provider is at. Our contribution to the literature and practice is on several fronts. 3 - Empowering Patients can Increase Digital Divide: The Effects of Patient Portals on Kidney Allocation YeongIn Kim, University of Texas-Dallas,Richardson, TX, 75080, United States, Mehmet Ayvaci, Srinivasan Raghunathan, Bekir Tanriover The recent healthcare reform promotes the use of information technology, such as patient portals, to provide patients better access to information sources. Motivated by the kidney transplant decision, we empirically analyze the impact of the patient portals on outcomes including time to deceased-donor transplant. We show that, overall, the adoption of a patient portal is positively associated with the probability of receiving a deceased-donor kidney. However, the impact is less for minority groups who have limited access to transplant service, which may imply further service divide in kidney transplant.

199