2015 Informs Annual Meeting

WE04

INFORMS Philadelphia – 2015

WE04 04-Room 304, Marriott Inventory Management - Stochastic Demand Contributed Session Chair: Nicholas Leifker, St. John Fisher College, 3690 East Ave., Rochester, NY, 14618, United States of America, nleifker@sjfc.edu 1 - Managing Inventory for a Stochastic and a Deterministic Demand Stream Rob Basten, Eindhoven University of Technology, P.O. Box 513, Eindhoven, 5600MB, Netherlands, r.j.i.basten@tue.nl, Jennifer Ryan We consider a stock point for an item that observes two streams of demands. Our motivating example is the maintenance of capital assets. The low priority demand is observed before parts need to be ordered and thus exhibits perfect advance demand information (e.g., preventive maintenance), while the high priority demand is observed afterwards (e.g., corrective maintenance). We characterize the structure of the optimal inventory control policy and we propose a myopic heuristic policy. 2 - Percentile Threshold Policies for Inventory Problems with Partially Observed Markovian Demands Parisa Mansourifard, PhD Candidate, University of Southern California, 1820 E Del Mar Blvd., Pasadena, CA, 91107, United States of America, parisama@usc.edu, Tara Javidi, Bhaskar Krishnamachari We consider the case of partially observed demand in the context of a multi- period inventory problem with lost sales. We present an interesting class of policies with a percentile threshold (PT) structure which outperforms the myopic policy and performs close to the optimal policy. We derive the performance guarantee of PT policies and present the optimal PT policy with a reasonable performance guarantee. 3 - A One-Warehouse Multi-Retailer Inventory System with Non-Homogeneous Poisson Demand Christian Bohner, Technische Universität Mönchen, Arcisstr. 21, Munich, Germany, christian.bohner@tum.de, Stefan Minner Product lifecycles and demand seasonality are important characteristics of inventory systems. We extend the continuous review one-warehouse multi- retailer inventory problem to non-homogeneous Poisson demand. Using the unit-tracking approach, we find optimal time-dependent parameters for base- stock policies both for the warehouse and the retailers. A numerical study shows that the exact dynamic solution clearly outperforms the solution obtained from time decomposition. 4 - Inventory Control of Intermittent Demand Combined with Economic Indicators Meng Yang, Tsinghua University, 519 Shunde Building, Beijing, 100084, China, yangm0628@gmail.com, Wanshan Zhu The inventory cost can be very high for expensive service parts of many companies, e.g., Caterpillar Inc., because their demand is highly unpredictable due to its intermittency. One way to reduce the cost is to make use of economic indicators that have a leading effect on the demand. We develop a Markov decision model to incorporate the economic indicator information for better controlling the inventory and quantify the value of this information. 5 - An Integrated Method of Optimization of the Final Order of Spare Parts Nicholas Leifker, St. John Fisher College, 3690 East Ave., At the end of a product’s life cycle, companies may place a final order of spare parts to satisfy all future demand for the part. Determining the optimal policy can be complicated when products contain multiple types of parts in which the failure rates of the parts and products are not independent; in such cases, the optimal final order quantities for all part types must be determined simultaneously. We explore the concavity properties of this optimization problem, and present a solution method. Rochester, NY, 14618, United States of America, nleifker@sjfc.edu, Timothy Lowe, Philip Jones

WE06 06-Room 306, Marriott Portfolio Analysis II Contributed Session

Chair: Dhanya Jothimani, Doctoral Student, Indian Institute of Technology Delhi, Department of Management Studies, New Delhi, India, dhanyajothimani@gmail.com 1 - The Robust Merton Problem of an Ambiguity Averse Investor Mustafa C. Pinar, Bilkent University, Faculty of Engineering, Ankara, Turkey, mustafap@bilkent.edu.tr, Sara Biagini We derive a closed form portfolio optimization rule for an CRRA investor diffident about mean return and volatility estimates. Confidence is represented by ellipsoidal uncertainty sets for the drift, given a volatility realization. The optimal policy is shaped by a rescaled market Sharpe ratio, computed under the worst case volatility. The result is based on a max-min HJB-Isaacs PDE, which extends the classical Merton problem and reverts to it for an ambiguity-neutral investor. 2 - An Orthogonal Genetic Algorithm for Indexing Tracking Problem Liang Bao, Professor, Xidian University, No. 2 South Taibai Road, Xi’An, China, baoliang@mail.xidian.edu.cn In this paper, we propose an orthogonal genetic algorithm for index tracking problem. Its significant feature is to incorporate an orthogonal design method into the initial population generation process and crossover operation. Our algorithm is more robust and can search the solution space in a statistically sound manner. We executed our algorithm to 5 datasets drawn from major world markets. The results compared with other published results show that our method has superior performance. 3 - Embedded Options in Institutional Investors’ Asset Allocation Problems Various options are embedded in institutional investors’ asset allocation problems. Pension funds are shorting a put option on the fund itself by requiring sponsors to contribute if underfunded. Sovereign wealth funds and family offices, have transfers from state or family businesses. The transfers depend on businesses’ performances and generate embedded options. We model the options and their implications on asset allocation with real option theory, stochastic control and dynamic programming. 4 - Modeling Uncertainties in Mean Variance Framework using Robust Optimization Dhanya Jothimani, Doctoral Student, Indian Institute of Technology Delhi, Department of Management Studies, New Delhi, India, dhanyajothimani@gmail.com, Ravi Shankar, Surendra Singh Yadav The classical mean variance (MV) framework ignores the uncertainties associated with the estimates of the expected returns; hence, the classical portfolio optimization problem is often called as error maximizer. In order to model the data uncertainty in MV framework, this study uses robust optimization technique to select the portfolios. The excess returns of portfolios obtained using robust estimators were found to be favorable compared to those obtained using classical estimators. Changle Lin, Princeton University, 10 Lawrence Drive, Apt 505, Princeton, NJ, 08540, United States of America, changlel@princeton.edu, John Mulvey

WE07 07-Room 307, Marriott Risk Analysis II Contributed Session

Chair: Maryam Tabibzadeh, California State University, Northridge, 1157 W., 30th St., Los Angeles, CA, United States of America, m.tabibzadeh@gmail.com 1 - Modular Production Capacity Expansion: An Examination of Collateral Risk Martin Wortman, Professor, Texas A&M University, Dept of ISEN, College Station, TX, 77843-3131, United States of America, wortman@atmu.edu, Cesar Malave Modular production operations are gaining considerable attention within electric power generation, chemical products, and bio-pharmaceutical industries. Modular capacity expansion can greatly reduce the financial risk associated with capitalizing production operations. However, modularized operations can also present collateral risk that can be greatly exacerbated. We offer an analytical explanation of this circumstance.

480