Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

WD24

3 - Robust Bounds of Default Probabilities in Financial Networks Dohyun Ahn, The Chinese University of Hong Kong, Shatin, N. T., Hong Kong We focus on firm-specific default probabilities in the Eisenberg-Noe network model with random shocks. In particular, we are interested in finding their robust bounds when financial networks are not fully observed. We first identify shock amplification caused by the network structure, which allows us to define the firm-specific default probabilities in financial networks. Using mixed-integer linear programming, we then obtain two kinds of robust bounds: one from a specific bank’s perspective and the other from regulators’ perspective. Also, we address asymptotic behaviors of those bounds when the shock size decreases. The applicability of those results is illustrated via numerical examples. 4 - Dynamic Programming, Progressive Hedging Algorithm and Machine Learning Xin Huang, PhD Student, The Chinese University of Hong Kong, Shatin, Hong Kong, Duan Li Dynamic programming often fails due to unavailability of analytical cost-to-go functions. In this research, we first solve the primary problem approximately by applying progressive hedging algorithm to associated scenario trees. Next we adopt neural networks to find mappings from states to values as approximated cost-to-go functions. We then update these approximated cost-to-go functions backward for all successive two stages by minimizing the gap between the prediction from preceding approximated function and that from the principle of optimality. This process iterates until achieving desired precision. n WD24 North Bldg 131B Practice- Productivity Improvement in Manufacturing Contributed Session Chair: Douglas Thomas, NIST, 100 Bureau Drive, Gaithersburg, MD, 20899-8603, United States 1 - The Direction of Causality between Personnel Management Practices and Productivity Juha Eskelinen, Researcher, Aalto University, Runeberginkatu 22-24, Helsinki, 00100, Finland, Ossi Aura, Guy Ahonen, Timo Kuosmanen Causal relationship between personnel management practices and productivity remains ambiguous. This longitudinal study combines survey data from Finnish small and medium size enterprises in manufacturing with productivity indicators for years 2008-2016 (17100 observations). Controlling for firm specific factors such as capital intensity, we find that measures regarding management development, work environment and content are associated with increase in future productivity. In contrast, measures like increasing non-financial employee benefits follow high productivity and do not precede it. 2 - The Impact of Patents on Manufacturing Productivity Anand Kandaswamy, Economist, NIST, Washington, DC, 20008, United States The traditional economics models in manufacturing usually look at factors like plant and equipment spending, worker education, and research and development funding to measure changes in productivity levels. What they tend to ignore is the role of innovation in driving productivity. Patents have traditionally been recognized as an excellent proxy for levels of innovation. Using data from 2004 to 2014, the author focused on six key manufacturing industries and created a model in R Studio that tries to determine the relative importance of innovation (as represented by patent activity) with respect to the more traditionally accepted components of economic productivity. 3 - The Effect of Flow Time on Productivity and Production Douglas Thomas, Economist, NIST, Gaithersburg, MD, 20899- 8603, United States Operations management strategies incorporate flow time and inventory turns as a metric for tracking and improving performance; however, there is limited understanding regarding the impact of reducing flow time. This paper examines the impact that flow time has on productivity and production, measured using the multifactor productivity index and manufacturing value added. A total of 6 models are presented and two simulations. The results suggest that flow time reflects and impacts productivity and production both within an industry and between industries. Moreover, the results are consistent with using flow time as a metric to identify industry level bottlenecks and improve productivity. 4 - Experimental Investigation of Stratasys J750 Polyjet Printer: Effects of Printing Parameters on Bending Strength Xingjian Wei, Graduate Student, Texas A&M University, College Station, TX, 77843, United States, Abhinav Bhardwaj, Zhijian Pei, Bruce Tai Additive Manufacturing (AM) has found significant applications in the engineering of prototypes as well as functional parts. These advances have resulted in the origination of high-resolution, multi-material printers such as the Stratasys J750: world’s first full-color, multi-material 3D printer. In this research,

we analyze the effects of printing parameters on the bending strength of samples printed by the Stratasys J750 printer. The results from this study would be valuable to research institutions and businesses leading high-resolution, multi- material 3D printing of polymers. 5 - The Motivation of Capital-giving in Crowdfunding Market: A Self-determination Theory Perspective Xiang Yuan, TongJi University, Shanghai, China, Hongwei Wang, Yuan Meng How to promote crowd-funding results successfully are crucial to crowdfunding projects. The results of crowd-funding projects are determined by investors’subjective behavior, which is triggered by some certain motivations. Through identifying different motivation modes mainly influenced by the project description, it will be beneficial to identify the investment intention of each investor. In this paper, based on the self-determination theory, we first create the corpus targeting different motives by means of the text mining method. Then, we classify the project description and project investment options. Last, we conduct an econometric model to examine the effect of investor’s motives. n WD25 North Bldg 131C Service Inventory Management Sponsored: Service Science Sponsored Session Chair: Seongkyoon Jeong, Arizona State University, Tempe, AZ, United States 1 - Applying Time Buffer Management to Lean Service Design Xiaofeng Zhao, University of Mary Washington, Fredericksburg, VA, 22406-7272, United States Theory of constraint-based buffer management is widely used to protect the system’s output from disruptions. The research employs queuing methods to calculate the time buffer size in lean service operation. It discusses computable expressions for the mean and the variation of waiting time. The approach can be implemented to conduct what-if analysis. 2 - Research on Supplier Evaluation in Service Procurement Based on Fuzzy Soft Set Jie Xu, Pittburgh, PA, 15213, United States The use of fuzzy soft sets can flexibly deal with multiple projects, different indicators, and uncertain service procurement items. Considering the experts’ case evaluation of the supplier’s various indicators, and based on the deterministic evaluation when considering the expert rating, the supplier’s negative (risk) factors and hesitation factors in each index are considered. Combining the expert’s affirmative scoring, negative and hesitant scores on supplier indicators, using triangular fuzzy numbers to represent the experts’ determination and negation of suppliers, and establishing an improved triangular fuzzy soft-package service procurement supplier evaluation model. 3 - Optimal Inventory and Redesign Strategies in Resolving Part Obsolescence Zhenyang Shi, Student, Shanghai Jiao Tong University, 1954 Huanshan Road, Shanghai, 200030, China, Shaoxuan Liu We study the joint part inventory and product redesign decisions for a manufacturer facing part obsolescence problem. Product redesign, and inventory control approaches such as last time buy, are two focal strategies in obsolescence resolution. We establish an optimal stopping model with additional decisions to investigate when should product redesign be initiated and how should part procurement and inventory be managed, during product life cycle after obsolescence. We show the optimality of a threshold policy for redesign choice and well-structured policies including the target interval policy for inventory control. 4 - Understanding Service Innovation Introduction and Diffusion Unlike tangible manufactured products, services have unique characteristics, which motivate the need to understand how service innovations are introduced and how they diffuse or percolate targeted markets. Using novel data of daily usage of online game services in South Korea, we find that service innovations in such online game industry appear to follow the saw-tooth patterns typically observed in the traditional inventory profile model of economic order quantity. We address what types of service products or innovations experience this phenomenon and what determines this type of introduction and diffusion profile. through the Lens of the Economic Order Quantity Model Seongkyoon Jeong, Arizona State University, Tempe, AZ, United States, Adegoke Oke

502