Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

MC22

3 - Online Platform Designs under Networked Cournot Competition John Pang, California Institute of Technology, Pasadena, CA, United States, Weixuan Lin, Eilyan Bitar, Adam Wierman We analyze designs of networked Cournot platform markets, considering (i) open access platform with transparent interactions to all participants, (ii) controlled allocation platform distributing aggregate productions efficiently, and (iii) discriminatory access platform allowing efficient interactions. We show that open access retains 2/3 of the optimal social welfare under anarchy while a controlled allocation platform optimizing social welfare can lead to anticipatory curtailment of production and inefficiency. Lastly, designing the network can lead to a 3/4 rate of optimal social welfare. We also introduce search cost, revealing the potential realistic inefficiency of open access. 4 - A Supply Chain Network Game Theory Model of Cybersecurity Investments with Nonlinear Budget Constraints Anna B. Nagurney, University of Massachusetts Amherst, Isenberg School of Management, Dept of Operations & Information Mgmt, Amherst, MA, 01003, United States, Patrizia Daniele, Shivani Shukla In this paper, we develop a supply chain network game theory model consisting of retailers and demand markets with retailers competing noncooperatively in order to maximize their expected profits by determining their optimal product transactions as well as cybersecurity investments subject to nonlinear budget constraints that include the cybersecurity investment cost functions. We provide alternative variational inequality formulations of the governing Nash equilibrium conditions, conduct Lagrange analysis, and apply the proposed algorithm to compute solutions to a spectrum of numerical supply chain network cybersecurity investment examples. 5 - Social Learning in a Dynamic Environment Agents learn about a moving state using private signals and past actions of their neighbors in a network. Can learning keep up with the changing environment? If private signals are sufficiently diverse and degrees sufficiently large, then Bayesian learning does very well, but not otherwise. Non-Bayesian learning rules do much worse, in contrast to environments with a fixed state. Stationary equilibria of Bayesian learning are characterized by linear rules reminiscent of the simple DeGroot heuristic. The resulting tractability can facilitate structural estimation of equilibrium learning models and testing against behavioral alternatives, as well as the analysis of welfare and influence. n MC22 North Bldg 130 Strategic Environments and Incomplete Information Sponsored: Revenue Management & Pricing Sponsored Session Chair: Yonatan Gur, Stanford University, Stanford, CA, 94305, United States Co-Chair: Daniela Saban, Stanford, New York, NY, 10027, United States 1 - Procurement in a Strategic Environment Gregory Macnamara, Stanford University, Stanford, CA, United States, Daniela Saban, Yonatan Gur We study a dynamic game of incomplete information that models the interactions between a Principal (“Buyerö), who demands the same good or service repeatedly over time, and an Agent (“Sellerö), who has private information. We characterize the effect that the strategic environment has on the Buyer’s ability to learn. 2 - Manufacturer Encroachment in a Non-exclusive Reselling Channel Parshuram Sambhajirao Hotkar, Doctoral Student, University of Texas at Austin, Austin, TX, 78751, United States, Stephen M. Gilbert We consider the implications of a manufacturer operating a direct channel to encroach upon the market of a non-exclusive reseller. Although non-exclusive resellers are common in practice, most existing studies of encroachment ignore the possibility of competing manufacturers selling through the same reseller. As we show, the presence of competing manufacturers has dramatic implications for how / when a direct channel should be used and who benefits from it. 3 - Team Decision Making in Operations Management Jiawei Li, PhD Candidate, Stephen M. Ross School of Business, 701 Tappan St, Ann Arbor, MI, 48109, United States, Stephen Leider, Damian Beil Existing behavioral OM literature has primarily studied individual decision makers. However, the behavioral economics literature suggests that teams may make better decisions in tactical settings, and may be more strategic and self- interested. We conduct a laboratory experiment and find that teams actually Nir Hak, Harvard University, Cambridge, MA, 02138, United States, Krishna Dasaratha, Benjamin Golub

perform worse than individuals when making Newsvendor decisions, exhibiting a stronger pull-to-center bias. In an information sharing game teams are less trustworthy when sharing information, but just as trusting when receiving information. Chat analyses are used to study the team decision-making process. 4 - Learning to Rank an Assortment of Products Shreyas Sekar, Harvard University, Cambridge, MA, United States, Kris Ferreira We consider the product ranking challenge that online retailers face when their customers typically do not have a good idea of the product assortment offered. These customers form an impression of the assortment after looking only at products ranked in the initial positions, and then decide whether they want to continue browsing all products or leave the site. We propose an online algorithm that learns consumer preferences and converges to the optimal full-information ranking.

n MC23 North Bldg 131A Financial Services Award Session

Sponsored: Finance Sponsored Session Chair: Agostino Capponi, Columbia University, 500 W 120th street, New York, NY, 10027, United States 1 - Addressing Systemic Risk Using Contingent Convertible Debt – A Network Analysis Lu Yueliang, UNC Charlotte, Charlotte, NC, USA. We construct a balance sheet based network model to study the interconnectedness of US banking system. After a simulation analysis of the buffer effect of contingent convertible {CoCo} debt in controlling contagion in the banking network under a theoretically motivated model, we use 13-F filings made to the US Securities and Exchange Commission (SED) to calibrate the theoretical model. While CoCo debt with dual-trigger is not so efficient as a single trigger design in reducing individual bank failures, the former is more effective at protecting the surviving banks, which leads to improved stability from the perspective of the banking system. We claim that the different designs of CoCo triggers result in trade-offs for addressing systemic risk, which may be evaluated in a network model. 2 - Beating the Curse of Dimensionality in Options Pricing and Optimal Stopping Yilun Chen, David Goldberg, Cornell University, Ithaca, NY, USA. The fundamental problem of pricing high-dimensional path-dependent options is central to applied probability and financial engineering, yet has remained open for the better part of a century. Modern approaches,often relying on approximate dynamic programming and/or the martingale duality theory for optimal stopping, either have limited rigorous performance guarantees, or have guarantees which scale poorly and/orrequire previous knowledge of good basis functions. A key di culty is that many approaches rely on computing nested conditional expectations, where the depth of nesting scales with the time horizon. 3 - Demand-Side Story of the Great Recession Marco Zhang, The University of Chicago Booth School of Business, Chicago, IL, USA. Using novel data on private pension plans’ investment allocations and returns, we find that the skewed interest for pension fund managers led to the increased investments in Mortgage-Backed Security (MBS) in the years preceding the Great Recession, especially for large pension plans with actuarial value over $10 billion. Such pension plans control much more investments per plan (they only represent less than 6% in our data but their aggregated investments is about 50% more than the remaining 94% combined) and are more likely to be under-funded (almost 64% of these plans are under-funded, compared to about 6% under- funded for the remaining plans). The skewed interest, together with the size of the investments the fund managers control, created both the will and the ability to dominate the MBS market. We find in our data that US private pension funds alone, which generally are much smaller compared to US public pension plans, put over $1 trillion in MBS-related investments in 2007 when the entire MBS market was about $9 trillion. The significant demand from the pension plans shifted the demand curve in the mortgage securitization market during the years prior to the Great Recession, which is supported by our observation of the concurrent increase of the quantity and the relative price for the MBS during the period. While there exists plenty literature studying the causes of the Great Recession from the supply-side, we are among the first to investigate and provide evidence of the roles pension plans played from the demand-side that contributed to the Great Recession.

192