Informs Annual Meeting 2017

MD28

INFORMS Houston – 2017

3 - Optimal Bidding Strategy for Market Participants Considering Both Renewable Generation and Price Uncertainties Kai Pan, University of Florida, 411 Weil Hall, Gainesville, FL, 32611, United States, kpan@ufl.edu, Yongpei Guan We present an optimal bidding strategy which is adaptive for the independent power producers to participate in both day-ahead and real-time markets by considering both renewable generation and price uncertainties. This proposed strategy is theoretically justified of its significant advantages over existing alternative ones. To improve the computational efficiency, polyhedral structures have been explored to strengthen the corresponding deterministic equivalent formulation. 4 - Economic Impacts Caused by the Diffusion of Hydrogen Energy Tetsuya Tamaki, assistant professor, Kyushu University, This research focuses on the economic impacts of the diffusion of hydrogen energy, which is an environment-friendly source of energy. Hydrogen energy is expected to reduce CO2 emission from not only the transportation sector but also the energy sector. However, the related technology is still expensive, and the usage and effects of hydrogen energy will change by the difference of the cost. In this research, to estimate the economic impacts caused by the diffusion of hydrogen energy, we use a dynamic optimization model that analyzes the economics related to climate change and also prescribe a policy for the use of hydrogen energy in society. 350B MAS Special Speaker: Mission-based Forecasting for Military Operations Sponsored: Military Applications Sponsored Session Chair: Greg H Parlier, North Carolina State University, 255 Avian Lane, Madison, AL, 35758, United States, gparlier@knology.net 1 - Mission-based Forecasting for Military Operations Greg H. Parlier, North Carolina State University, 255 Avian Lane, Madison, AL, 35758, United States, gparlier@knology.net The US Department of Defense operates the world’s most complex global supply chains. To address persisting problems the US Army established the project to Transform Army Supply Chains (TASC) in order to investigate the nature, causes, and consequences of demand uncertainty and supply variability. Mission-Based Forecasting is a new concept for demand planning which relates resource investment levels and distribution policies directly to mission performance outcomes, thereby enabling the “resources-to-readiness” linkage. MBF improves forecast accuracy and reduces backorders, excess inventory, and costly work- arounds while increasing equipment readiness in military organizations. 350C Mining Text and Graph Data from Online Platforms Invited: Social Media Analytics Invited Session Chair: Theodoros Lappas, Stevens Institute of Technology, 310 10th St, Hoboken, NJ, 07030, United States, tlappas@stevens.edu 1 - Social Connections and Venture Investments Chuanren Liu, Drexel University, Philadelphia, PA, United States, chuanren.liu@drexel.edu We present a Social-Adjusted Probabilistic Matrix Factorization (PMF) model to exploit the social relationship information from VC firms and startups for startup investment recommendations. Unlike previous studies, we make use of the directed social connections and quantify the variety of social network groups. As a result, this brings in much more flexibility, and the modeling results inherently provide meaningful managerial implications for the operators of VC firms and start-up companies. We evaluate our model on both synthetic and real-world data. The results show that our approach outperforms the baseline algorithms with a significant margin. 2 - Measuring Reputation Inflation in Online Platforms Apostolos Filippas, New York University, New York, NY, United States, afilippa@stern.nyu.edu In many online marketplaces, the distribution of feedback scores seems implausibly rosy. Using data from a large online marketplace, we show that average feedback scores have risen sharply over time, leading to substantial top- censoring. However, the increase in reputation could be either due to more satisfied raters, or raters who are lowering their standards. We employ a text- MD26 MD27 744 Motooka, Fukuoka City, 8190395, Japan, tamaki@doc.kyushu-u.ac.jp, Shunsuke Managi

analytic approach to disentangle the two reasons, and we show that 36 to 48 percent of this increase is due to a phenomenon akin to inflation. We discuss the implications for our results for online marketplace design. 3 - Text-based Industry Classification and Network Feng Mai, Stevens Institute of Technology, Hoboken, NJ, United States, feng.mai@stevens.edu, Lei Zheng Traditionally, firms are placed within predefined industry groups using SIC or NAICS codes. These classifications are time-invariant and are not robust to changing primary business activities. We propose and validate a new text-based method that can quantify how firms differ from their competitors and redefine industry boundaries. 4 - Arousal, Hedonism, and Utility in Online Reviews Jie Ren, Fordham University, New York, NY, United States, jren11@fordham.edu, Jeffrey V. Nickerson To explain sales, in addition to online review valence and volume, this paper bases its argument on one less-studied hedonic versus utilitarian dimension of product type and also on a less-studied online review dimension - arousal, a measure that is extracted from individual review texts. Using panel data analysis on 26,357 Amazon products and an online experiment involving 600 subjects, the findings consistently show that product type moderates the impact of the three dimensions of online reviews (valence, volume, and arousal) on sales. These findings explain why sometimes online review valence is more influential than volume with respect to sales and why sometimes it is the other way around. Invited: Auctions Invited Session Chair: Umut Dur, NCSU, Raleigh, NC, 27695, United States, umutdur@gmail.com 1 - Parallel versus Sequential School Admissions: A Tale of Two Countries Onur Kesten, Associate Professor of Economics, Carnegie Mellon University, 5000 Forbes Avenue, POS.- Posner Hall - Room 249, Pittsburgh, PA, 15213, United States, okesten@andrew.cmu.edu This paper studies the recent reforms in the student assignment systems in Turkey and Sweden. There are two types of high schools in both countries: public and private schools. Public school admissions are administered in a centralized manner, whereas private school admissions are done in a decentralized manner. In Turkey, starting from 2014, private school assignments are carried out in a first round along with a policy change in the terms re-entry into the second round. We study the performance of old and new sequential assignment systems in both countries and show that alternative solutions that outperform both systems are available. 2 - Matching with Limited Information: A Case for Neighborhood Schools Peter Troyan, Assistant Professor, University of Virginia, Charlottesville, VA, 22904, United States, pgt8y@virginia.edu, Andrew Kloosterman Deferred acceptance is a popular school choice mechanism because it is strategyproof and stable. However, this relies on the assumption that all agents know their own preferences. We relax this by allowing agents to be either informed or uninformed. Ex-post stable outcomes need not exist, as uninformed agents may form a blocking pair after learning their match and updating their beliefs. Further, we find that informed agents gain at the expense of uninformed agents, who are more likely to be left with underperforming schools. Positive results can be recovered by giving all students a ``neighborhood school’’: the equilibrium outcome is ex-post stable, and uninformed students are made better off. 3 - Sequential School Choice Theory and Evidence from the Field and Lab Umut Dur, NCSU, Raleigh, NC, United States, umutdur@gmail.com, Robert Hammond, Thayer Morrill, Onur Kesten We analyze sequential preference submission in matching problems. Our motivation is school districts that use a website where students submit preferences sequentially. Comparing the Boston Mechanism (BM) to the Deferred Acceptance (DA), we show that sequential BM is more efficient than sequential DA. Any equilibrium outcome under BM Pareto dominates the student optimal stable matching. We present two sets of empirical tests. First, we study a field setting in which sequential BM was used. The field data provide evidence that is consistent with our theory. Second, we conduct an experiment to do comparison in a lab. BM improves upon DA when students submit sequentially but not simultaneously. MD28 350D Matching Theory

245