2015 Informs Annual Meeting

MD06

INFORMS Philadelphia – 2015

MD04 04-Room 304, Marriott Panel: International Opportunities Sponsor: Junior Faculty Interest Group Sponsored Session Chair: Raha Akhavan-Tabatabaei, Associate Professor, Universidad de los Andes, Carrera 1 Este # 19 A - 40, Bogota, Colombia, r.akhavan@uniandes.edu.co Co-Chair: Shengfan Zhang, Assistant Professor, University of Arkansas, 4207 Bell Engineering Center, Fayetteville, United States of America, shengfan@uark.edu 1 - A Panel Discussion on International Opportunities Moderator:Shengfan Zhang, Assistant Professor, University of Arkansas, 4207 Bell Engineering Center, Fayetteville, AR, United States of America, shengfan@uark.edu, Panelists: Dionne Aleman, Andres Medaglia, Fugee Tsung This panel consists of department heads, junior and senior faculty members of universities in Canada, Colombia, Hong Kong and Turkey. They will discuss opportunities in academic jobs, tenure and promotion processes, research resources (funding and visiting opportunities), professional development, recruiting, etc. MD05 05-Room 305, Marriott Predicting Customer Behavior using Facebook Data Cluster: Social Media Analytics Invited Session Chair: Michel Ballings, Assistant Professor Of Business Analytics, The University of Tennessee, 249 Stokely Management Center, Knoxville, TN, 37996, United States of America, michel.ballings@utk.edu 1 - Using Customers’ Facebook Pages to Improve Lead Qualification in a B2B Acquisition Process Matthijs Meire, PhD Student, Ghent University, Tweekerkenstraat, 2, Ghent, 9000, Belgium, Matthijs.Meire@ugent.be, Michel Ballings, Dirk Van Den Poel The purpose of this study is to investigate the added value of Facebook data in B2B customer acquisition. We use a Random Forest prediction model. The results indicate that adding customers’ Facebook page data can indeed improve B2B lead qualification. Our contribution is twofold. First, to the best of our knowledge it is the first to use Facebook data in B2B lead qualification. Second, we quantify the monetary gains of using Facebook data by conducting a real-life lead targeting experiment. 2 - Investigating the Drivers of Likes and Comments on Facebook Steven Hoornaert, PhD Student, Ghent University, Ghent, 9000, Belgium, Steven.Hoornaert@ugent.be, Michel Ballings, Dirk Van Den Poel The objective of this study is to investigate the added value of user context data in Facebook post popularity prediction models. For this purpose, two Random Forest models were built: one including only post variables (e.g., post type) and another containing both post and user variables (e.g., age). Predictability is improved for likes (x3.7) and comments (x3.6). This study is the first to augment post popularity prediction models with user context data and analyze a large quantity of posts. 3 - Predicting Buyer Behavior using Social Media Data Matthias Bogaert, PhD Student, Ghent University, Tweekerkenstraat 2, Ghent, 9000, Belgium, matthias.bogaert@ugent.be, Michel Ballings, Dirk Van Den Poel The purpose of this study is to explain customer behavior (offline event attendance) based on SM data. In order to substantiate our findings, we used propensity score matching and built a Random Forest model. This study reveals that social media data can predict offline event attendance with high predictive accuracy. Moreover, the results suggest that the number of friends that are attending the focal event and event attendance on Facebook were highly significant.

4 - The Power of Facebook to Predict Customer Acquisition and Defection Michel Ballings, Assistant Professor Of Business Analytics, The University of Tennessee, 249 Stokely Management Center, Knoxville, TN, 37996, United States of America, michel.ballings@utk.edu, Matthijs Meire, Dirk Van Den Poel The main purpose of this study is to investigate the value of Facebook data in predicting individual customer behavior. In addition we study the importance of different online engagement variables such as likes, answers to event rsvp’s, and group memberships in predicting acquisition and defection. The results indicate that customer acquisition can be predicted very accurately using Facebook data. In addition Facebook data significantly improve defection prediction over and above customer data.

MD06 06-Room 306, Marriott Finance and Risk Management Sponsor: Financial Services Sponsored Session

Chair: Samim Ghamami, Board of Governors of the Federal Reserve System, 20th Street and Constitution Avenue N.W., Washington, DC, United States of America, samim.ghamami@frb.gov 1 - Derivatives Pricing under Bilateral Counterparty Risk Samim Ghamami, Board of Governors of the Federal Reserve System, 20th Street and Constitution Avenue N.W., Washington, DC, United States of America, samim.ghamami@frb.gov We consider risk-neutral valuation of a contingent claim under bilateral counterparty risk in a setting similar to that of Duffie and Singleton (1999). We develop probabilistic valuation formulas that have closed-form solution or can lead to computationally efficient pricing schemes. Drawing upon the work of Ghamami and Goldberg (2014), we show that derivatives values under wrong way risk (WWR) need not be less than the derivatives values in the absence of WWR. 2 - Stochastic Intensity Margin Modeling of Credit Default Swap Portfolios Dong Hwan Oh, Economist, Federal Reserve Board, 20th Street and Constitution Avenue N.W., Washington, DC, 20551, United States of America, donghwan.oh@frb.gov, Samim Ghamami, Baeho Kim We consider the problem of initial margin (IM) modeling for portfolios of credit default swaps (CDS) from the perspective of a derivatives CCP. Inspired by Cont and Kan (2011), the CCPs’ IM models in practice are based on theoretically- unfounded direct statistical modeling of CDS spreads. Using the well-known reduced-form approach, our IM model prices the portfolio constituents in a theoretically meaningful way and shows that statistical IM models can underestimate CCPs collateral requirements. 3 - Evaluating Central Counterparty Risk Anton Badev, Economist, Federal Reserve Board, 1801 K St. NW, Washington, DC, 20006, United States of America, anton.i.badev@frb.gov, Samim Ghamami A conceptually sound and logically consistent definition of the CCP risk capital is challenging, and incoherent CCP risk capital requirements may create an obscure environment. Based on novel applications of well-known mathematical models in finance, this paper introduces a risk measurement framework that coherently specifies all layers of the default waterfall resources of typical derivatives CCPs. We apply the proposed framework on DTCC data and evaluate various risk management practices. 4 - Risk Screening in Microfinance: Modeling and an Extragradient- based Online Learning Algorithm Yuqian Xu, NYU Stern School of Business, 44 West 4th Street,

New York City, NY, 10002, United States of America, yxu@stern.nyu.edu, Michael Pinedo, Binqing Xiao

In this paper, we get the business loan application and default data from one of the leading banks in China. We then propose a statistical model with three different types of indexes to quantify the potential performance of a firm: its financial level index, operational level index, and business owner level index and provide an efficient extragradient-based online learning algorithm to solve it.

233