Informs Annual Meeting 2017

MA16

INFORMS Houston – 2017

MA16

MA17

332F Inventory Management Contributed Session

340A Quantitative Trading and Optimal Execution Sponsored: Applied Probability Sponsored Session Chair: Ciamac Cyrus Moallemi, Columbia University, New York, NY, 10027, United States, ciamac@gsb.columbia.edu Chair: Costis Maglaras, Columbia University, New York, NY, 10027, United States, c.maglaras@columbia.edu 1 - Microfoundations of Multifactor Model for Trading Volume in Speculative Market Seungki Min, PhD Candidate, Columbia University, 1 16th St Broadway, New York, NY, 10027, United States, smin20@gsb.columbia.edu The joint distribution of trading volume across multiple securities, as a stochastic model of liquidity, needs to be considered in the execution of portfolio transactions. We propose a microstructural model that explains trading volume with the investors’ activity responding to sequential information arrival. Only with idiosyncratic investors, our model restores ‘Market Microstructure Invariance’ hypothesis of Kyle and Obizhaeva (2016). With factor portfolio investors, it suggests a factor structure in trading volume which we observe empirically for U.S. equities. We also observe the increase in factor investment during the recent years that implies the growth of passive investing. 2 - A Separation Principle for Dynamic Portfolio Optimization Ciamac Cyrus Moallemi, Columbia University, Columbia Business School, 3022 Broadway Uris 416, New York, NY, 10027, United States, ciamac@gsb.columbia.edu, Benjamin Van Roy We consider multi-period portfolio optimization problems in the presence of predictablereturns. In our setting, we establish that that the term structure of future conditional meanreturns is a sufficient statistic for optimal decision making. This provides a separationprinciple that partitions optimal trading into two natural steps: (1) forecast the term structureof future expected returns; (2) compute the optimal trading decision by solving a single-period,deterministic Markowitz-style mean-variance optimization problem, but with an endogenouslydetermined effective time horizon. 3 - Robust Wasserstein Profile Inference in Distributionally Robust Mean-variance Model Lin Chen, Columbia University, New York, NY, 10025, United States, lc3110@columbia.edu, Jose Blanchet, Xunyu Zhou This project is concerned with a distributionally robust mean-variance model. By solving its dual problem, we can get a data driven model with penalty on norm of portfolio. We use RWPI method to choose the regularization parameter without the use of cross validation and then design an algorithm to solve the optimization problem. Numerically experiments are given to compare the performance of portfolio obtained by traditional Markovitz model and RWPI method. 340B Stochastics and Behavioral Operations Sponsored: Applied Probability Sponsored Session Chair: Yash Kanoria, Columbia Business School, New York, NY, 10027, United States, ykanoria@gmail.com 1 - Traffic Scheduling:Large Population Optimality Harsha Honnappa, Purdue University, 315 N. Grant Street, West Lafayette, IN, 47906, United States, honnappa@gmail.com We solve the problem of scheduling appointments for a finite customer population to a service facility with customer no-shows to minimize customer waiting time and server overtime costs. We study the problem in fluid and diffusion frameworks. We find that it is optimal to schedule traffic so that the system is in critical load. That is, heavy-traffic is obtained as a result of optimization rather than as an assumption. Our proposed schedules are proved to be asymptotically optimal for a finite horizon in the fluid scaling and for large and finite horizon in the diffusion scaling. We also provide some initial results on solving the finite horizon problem. MA18

Chair: Stef Lemmens, KU Leuven, Naamsestraat 69 Box 3555, VAT BE 0419.052.173, Leuven, 3000, Belgium, stef.lemmens@kuleuven.be 1 - Joint Replenishment Problem with Batch Ordering Ana Muriel, University of Massachusetts Amherst, Dept of Mechanical & Industrial Eng, 160 Governors Drive, Amherst, MA, 01003, United States, muriel@ecs.umass.edu, Michael Prokle Motivated by industrial partners who are exploring the benefit of coordinating the ordering and shipping of a variety of parts sold in boxes of a given size, we study the JRP with batch ordering. Given a constant reorder interval, we determine the length of a regeneration period and derive a closed-form expression for the average inventory. We find the optimal constant reorder interval and show the savings relative to an uncoordinated strategy and the joint EOQ policy. The optimal ordering policy under constant demand, however, requires a non-constant reorder interval. We propose a MIP to determine the optimal policy and explore the relative quality of constant reorder interval policies. 2 - The Impact of Political Preference on Inventory Decisions Yi Tan, University of Kansas, 1654 Naismith Drive, Lawrence, KS, 66045, United States, vieiraty0113@gmail.com, Yaoyi Xi We conduct an empirical investigation of whether and how CEO political ideology affects firm inventory strategies. Using a panel that contains quarterly data from 1991 to 2016 for 2,745 unique U.S. firms covered by Compustat, we find that firms tend to hold more inventories when the Party their CEOs prefer holds the Presidency. We perform further analyses on firm financial performance to Stephen M.Disney, Professor of Operations Management, Cardiff Business School, Aberconway Building, Colum Drive, Cardiff, CF10 3EU, United Kingdom, disneysm@cardiff.ac.uk, Robert Boute, Jan A.Van Mieghem Companies often experience non-stationary demand as products evolve over their life cycle. We investigate a tractable family of single and dual sourcing policies tailored to such demand environment. We adopt a conventional discrete time inventory model with a linear control rule that smooths orders and allows an exact analytical analysis of an easy to implement dual sourcing policy. 4 - Not Centralizing your Inventory? It may be Fine Bo Li, Assistant Professor, Ningbo Supply Chain Innovation Institute China, Ningbo, China, libo@alum.mit.edu, Antonio Arreola-Risa, Mark L. Spearman Centralizing inventory, especially safety stocks, is often a proposed solution to cut inventory. In this talk, we consider a two-tier production-inventory system and illustrate how the benefit of centralizing inventory can be negligible under certain situations. 5 - Performance Measurement of a Rotavirus Vaccine Supply Chain by Integrating Production Capacity into the Guaranteed Service Approach Stef Lemmens, KU. Leuven, Naamsestraat 69 Box 3555, VAT.BE 0419.052.173, Leuven, 3000, Belgium, stef.lemmens@kuleuven.be, Nico Vandaele, Catherine Jenny Decouttere, Mauro Bernuzzi, Amir Reichman Previous research has integrated multi-echelon inventory management into the design of a responsive supply chain by the use of the guaranteed service approach. We model the production capacity with a queuing network to handle the variability of the batch production processes as well as the demand variability. The focal point is the tail of the lead time distribution which has an impact on the supply chain safety stock. We apply our model to the rotavirus vaccine supply chain. The impact of modelling capacity on lead times and thus safety stocks and working capital is significant. investigate the underlying mechanisms for this finding. 3 - Global Dual Sourcing and Order Smoothing

138