Informs Annual Meeting 2017

WC55

INFORMS Houston – 2017

2 - Sparse Travel Time Estimation from Streaming Data Deepthi Mary Dilip, New York University-Abu Dhabi, Abu Dhabi, United Arab Emirates, deepthi.dilip@nyu.edu, Saif Eddin G. Jabari We address two shortcomings in travel time distribution estimation methods. The first is the determination of the number of modes of the distribution. The second is the use of Gaussian kernels, which can assign positive probabilities to negative travel times. We develop a sparse kernel density estimation technique using asymmetric kernels. Sparsity promotes parsimony and helps avoid the need for a predefined number of mixture components. We present an adaptive approach to infer both the location and the scale parameters of the kernels and we propose a generalization of Gamma kernels. 3 - Analytical Calculation of Travel Time Reliability Indices with Minimal Statistical Data Marcelo Ricardo Figueroa, Rutgers University, 93 Marvin Lane, Piscataway, NJ, 08854, United States, marcelo.figueroa@rutgers.edu We present an analytical parametric model to approximate the travel time distribution of vehicles traversing a freeway corridor which experiences random quality of service degradations. Our models are validated by using traffic speed and flow data for a freeway corridor in Milwaukee, Wisconsin coupled with relevant weather events and traffic incidents as the main cause of service degradation. We show how the models can be calibrated using minimal statistical information on traffic conditions. We provide travel time (un)reliability measures for various scenarios illustrating the impact of incident duration, frequency, and degradation level. 4 - Statistical Inference of Travel Demand with Complex Mapping Yudi Yang, Postdoc Researcher, University of California, Davis, 1420 Lake Boulevard, Apartment 17, Davis, CA, 95616, United States, ydyang@ucdavis.edu, Yueyue Fan, Johannes Royset We study a novel problem of inferring the joint probability density function (pdf) of O-D demand variables by integrating hard data of observed traffic counts and potential soft information on demand features. Without being restricted by any particular distributional assumption, our moment matching method produces a nonparametric epi-spline density estimate. Based on sampling instead of solving inverse problems, the proposed solution process can be easily paralleled and is computationally tractable when a complex mapping is assumed between trip rates and traffic flow, including those based on optimization and simulation. Chair: Pei-Cheng Liao, National Taiwan University, No 1 Sec 4 Roosevelt Road, Taipei, 10617, Taiwan, pcliao@ntu.edu.tw 1 - The Usefulness of Forward-looking and Backward-looking Information in the Capital Market Yue Wang, University of Texas-Austin, 2110 Speedway, Stop B6400, Austin, TX, 78712, United States, yue.wang@mccombs.utexas.edu We investigate circumstances in which forward-looking or backward-looking financial information is more useful in the capital market. The choice between forward-looking and backward-looking information reflects a tradeoff between “relevancy” and “reliability.” We use conference call transcript to analyze the frequency of forward-look and backward-looking sentences, as it provides effective communication and interaction between managers and sell-side analysts. The results will be potentially helpful to regulators and preparers of financial statements who wish to provide useful information to investors. 2 - Official Visits and Corporate Employee Scale Wang Yanping, doctor, Tongji University, Yangpu District Zhangwu Road NO1 1822, Shanghai, China, wangyanpingchina@163.com Through the official visit statistics of Fortune 500 global corporations, we find that different from the developed market economy countries, the Chinese officials visit the corporation frequently, forming “Chinese Characteristics” visit phenomenon. As to employment issue, the empirical findings show that Chinese officials are inclined to visit the corporations which can absorb more employees. Chinese official visits can increase the number of employees, especially for employees with college degree or below, thus incurring redundant employee. This relation is much stronger in state-owned enterprises. 3 - The Effect of Top Managers’ Regulatory Focus on Firm Behaviors Jacob Do-Hyung Cha, Seoul National University, Room 313, 58-dong, Seoul National University, Gwanak-gu, 150-742, Korea, Republic of, research.dohyung@gmail.com, Hye-Jin Cho Individuals pursue goals via either a prevention focus (sensitive to gains and a desire for accomplishment and growth) or a promotion focus (sensitive to losses and a desire for safety and security). This study investigates whether individual managers’ such motivational traits are associated with corporate behavior and performance. I use linguistic features to develop a measure of managers’ WC55 362B Accounting Contributed Session

regulatory focus - prevention or promotion foci - from corporate documents, that is, Form 10-K’s management discussion and analysis section. The results indicate that managers’ regulatory focus affect differently financing choices, investment choices, and firm operating performance. 4 - The Stewardship Role of Conservatism Pei-Cheng Liao, Professor, National Taiwan University, No 1 Sec 4 Roosevelt Road, Taipei, 10617, Taiwan, pcliao@ntu.edu.tw, Guang Ma, Suresh Radhakrishnan We consider conservative reporting and investment decisions when the hidden- information problem of unknown manager-types as well as the hidden-action problem of diversion of resources exist. When either the hidden-information problem or the hidden-action problem exists, conservative reporting is not optimal. When both the hidden-information and hidden-action problems exist, conservative reporting is optimal for the high-type manager. Specifically, conservatism as part of the menu targeted for the higher productivity manager mitigates the problem of diversion of resources by the lower productivity manager and thus is part of the steward role of accounting. 362C Logistics Contributed Session Chair: Murat Karatas, University of Texas-Austin, Austin, TX, United States, mkaratas@utexas.edu 1 - Optimization of Supply Chain Tracking Capability under Various Market Contexts Bin Dai, Professor, Wuhan University, Bayi Road, Wuhan, China, dbbudstar@gmail.com, Yongzhe Zhang Traceability system has been widely used to alleviate the impact of product recall on supply chain performance, particularly the product recall cost. In this paper, we optimize the supply chain tracking capability under various market contexts such as oligopoly, simutaneouly move competition and sequential move competition, and then the optimal tracking capability and pricing strategies are compared. 2 - Benders Decomposition for Inventory Routing Problem with Perishable Product Faisal M.Alkaabneh, Cornell University, Stewart Ave, Building 516; Room, Ithaca, NY, 14850, United States, fma34@cornell.edu In this work we consider a Perishable products Inventory vehicle Routing Problem in the context of Cold supply chain. We explicitly model Green House Gases emissions due to transportation activities depending on the vehicle load during delivery. The problem is modelled as a mixed integer mathematical program that is difficult to solve directly. To this end, we developed a solution methodology based on Benders decomposition. Benders decomposition algorithm convergence rate was accelerated through the addition of valid inequalities and symmetry breaking cuts to the Master Problem, upper bound heuristic, Pareto optimal cuts and trust region. 3 - A Decomposition Heuristic for a Production Routing Problem in Furniture Companies Pedro Miranda, Federal University of São Carlos, São Carlos, Brazil, pmiranda@dep.ufscar.br, Jean-François Cordeau, Deisemara Ferreira, Raf Jans, Reinaldo Morabito We propose a decomposition heuristic to solve a production routing problem arising in the context of furniture companies. The heuristic decomposes the problem into two subproblems that are solved iteratively. The first subproblem makes production and distribution decisions using approximations of the routing costs and travel times. The second subproblem determines vehicle routes based on the decisions made by the first subproblem. These routes are used to update the approximations of routing costs and travel times for the next iteration. Computational results show that our iterative approach provides better solutions in less computing times than a state-of-the-art commercial solver. 4 - Multi-warehouse Package Consolidation for Split Order Fulfillment in Online Supermarkets Yuankai Zhang, Dalian University of Technology, Office Room A1214 Innovation Park Building, Dalian University of Technology No.2 Linggong, Dalian, 116024, China, ykzhang6635@126.com, Xiangpei Hu, Weihua Lin The split-order fulfillment problem has been a new key issue for multi-item order fulfillment in Chinese online supermarkets. We propose a powerful order fulfillment method - package consolidation - in which two or more stock keeping units in the sub-orders of one order will be packed into fewer packages. We formulate a multi-commodity network flow model to make the package consolidation decisions for split orders between multiple warehouses, and propose a computationally efficient algorithm for solving it. WC56

522