2016 INFORMS Annual Meeting Program

MB41

INFORMS Nashville – 2016

3 - Reducing Carbon Emissions In Grocery Retail Ekaterina Astashkina, INSEAD, Boulevard de Constance, Fontainebleau, 77305, France, ekaterina.astashkina@insead.edu, Elena Belavina We build a stylized model for traditional and online grocery retail chains to understand the drivers of the consumer and retailer carbon footprint, including emissions that come from food waste and transportation. In our model, consumers make endogenous choices between different channels and the associated food-buying policies, while retailers optimally manage their inventory replenishment. We find that, in most cases, the availability of an online retailer reduces the emissions associated with the grocery sector in a city. We also consider the effectiveness of alternate policy instruments including sales and carbon taxes, and identify actions that improve the behavior of the worst offenders. 4 - Optimizing Water Pollution Monitoring System: Regulation Policy Guideline For Curbing Nutrient Pollution Michael Lim, U of Illinois at Urbana-Champaign, Champaign, IL, 61820, United States, mlim@illinois.edu We examine regulatory guidelines of surface water quality to curb nutrient pollution resulting from various farming activities. Specifically, we formulate an optimization model that captures the government’s regulation decision taking into account farmers’ moral hazard issues: determining the optimal location of monitor stations along with optimal penalty schemes for each watershed district. We explore the model using the Illinois State water network to ensure practical relevance and to obtain further insights on regulation policy. MB38 206A-MCC Behavioral Modeling with Social Data Invited: Social Media Analytics Invited Session Chair: Tauhid Zaman, MIT, 77 Mass Ave, Boston, MA, 02139, United States, zlisto@mit.edu 1 - Optimal Policies For Finding Users Hiding In Social Networks Christopher Marks, MIT, cemarks@mit.edu During 2015 we collected data from approximately 5000 Twitter accounts belonging to ISIS users, ISIS supporters, and other users that appeared to be closely connected to the ISIS network. We observe that many of these users are frequently suspended, only to immediately open new accounts from which they continue their online activities. We present a dynamic search method for finding new accounts belonging to previously suspended users that relies on machine learning methods to generate model inputs. We analyze this search method in the context of dynamic programming and provide some insights into characteristics of an optimal search policy. 2 - Optimal Following Policies In Social Networks Using Integer Programming And Network Centrality Tauhid Zaman, Massachusetts Institute of Technology, zlisto@mit.edu, Krishnan Rajagopalan We consider the problem of interacting with users in a social network in order to maximize the number of followers obtained. We formulate the problem as an integer program (IP). We then show how to dramatically speed the time needed to solve the IP by modifying the objective using network centrality functions. Through simulations on real social networks, we find that our modified IP can increase the number of followers obtained versus random and pure network centrality based policies. 3 - Bayesian Inference Of User Geolocation Using Social Media Activity Time Series Matthew Robert Webb, MIT, Cambridge, MA, United States, mrwebb@mit.edu We propose a novel Bayesian classification algorithm to determine the global location of Muslim extremists from their social media activity based on their unique pattern of life. The tenants of Islam require five daily prayers; but rather than being set, prayer times are determined by the location of the Sun in relation to the Earth’s horizon. By assuming Muslim users will not utilize social media during prayers, we attempt to infer their longitude and latitude based on their pattern of inactivity. 4 - The Value Of Social Media To Online Content Michael Zhao, MIT, Cambridge, MA, United States, mfzhao@mit.edu, Sinan Aral Many believe social media drives online content consumption and vice versa. The potential of this positive feedback loop is critical to marketers, publishers, politicians, and beyond. However, this type of relationship induces endogeneity problems that make casual identification difficult. We overcome this challenge by

constructing a unique article-location panel dataset using proprietary data from a large online and print media company. We employ a novel IV estimation strategy by using location-specific weather patterns as instruments for social media sharing thereby allowing us to identify the degree to which social media effects the demand for online content. MB39 207A-MCC Panel: Future of Applied Probability Sponsored: Applied Probability Sponsored Session Chair: David Goldberg, GA Institue of Technology, Atlanta, GA, United States, dgoldberg9@isye.gatech.edu 1 - Future Of Applied Probability David Goldberg, GA Institue of Technology, 755 Ferst Drive, Atlanta, GA, 30332-0205, United States, dgoldberg9@isye.gatech.edu An opportunity for the entire Applied Probability Community to discuss the future of the field. 2 - Panelists Applied Probability Community, Applied Probability Community, INFORMS, Catonsville, MD, 21228, United States, meetings@informs.org Chair: Rafael Mendoza, McCombs School of Business, University of Texas, Austin, TX, 78712, United States, rafael.mendoza-arriaga@mccombs.utexas.edu 1 - On Latency And Volatility Richard Sowers, University of Illinois, r-sowers@illinois.edu We present a simple model of the effects of latency on the properties of observed asset prices. In our model, latency is a delay between the observed asset price and its true, but latent fundamental price. Because of latency, the observed asset price shadows the true but latent asset price at some deformed time away. Deformation in a clock gives rise to fluctuations in volatility. We provide an asymptotic result that links latency to the volatility of volatility. 2 - Energy Production & Games With Stochastic Demand Ronnie Sircar, Princeton, sircar@princeton.edu The dramatic decline in oil prices, from around $110 per barrel in June 2014 to around $30 in January 2016 highlights the importance of competition between different energy sources. Indeed, the price drop has been primarily attributed to OPEC’s strategic decision not to curb its oil production in the face of increased supply of shale gas and oil in the US, coupled with reduced demand from China. We model these phenomena as dynamic Cournot games in a stochastic demand environment, and illustrate how traditional oil producers may react in counter- intuitive ways in face of competition from alternative energy sources. 3 - Welfare Analysis Of Dark Pools Krishnamurthy Iyer, Cornell University, Ithaca, NY, United States, kriyer@cornell.edu, Ramesh Johari, Ciamac Cyrus Moallemi We investigate the welfare implications of operating alternative market structures known as “dark pools” alongside a “lit” dealer market. Our setting consists of intrinsic traders and speculators, with heterogeneous private information as to an asset’s value, who endogenously choose between the two venues. We establish that while the dark pool attracts relatively uninformed traders, the orders therein experience adverse selection. Moreover, the informational segmentation created by a dark pool leads to greater transaction costs in the lit market. From this, we conclude that there exist reasonable parameter regimes where the introduction of a dark pool decreases the overall welfare. MB41 207C-MCC Advances in Quantitative Finance Sponsored: Financial Services Sponsored Session

161