Latest Seminars

Effect of Consumer Awareness on Corporate Social Responsibility Under Asymmetric Information
Dr Xiaomeng Guo, Assistant Professor, Department of Logistics and Maritime Studies, Faculty of Business, The Hong Kong Polytechnic University

This paper studies the interaction between a firm and consumers under the consideration of corporate social responsibility. The firm can be either socially responsible or socially irresponsible; however, the consumers cannot observe the firm's exact type, which is private information. The firm can try to signal its type through pricing and other information-disclosure mechanisms (e.g., issue sustainability reports or third-party certifications). We find that due to the existence of asymmetric information, increasing consumer awareness of corporate social responsibility may not necessarily benefit a firm with socially responsible behaviors. More specifically, when a larger fraction of consumers become socially concerned or when the consumers have stronger willingness to reward (punish) the responsible (irresponsible) firm, the responsible firm could be worse off whereas the irresponsible firm could be better off. This is because the seemingly favorable trend in consumer behavior will affect the responsible firm's signaling cost as well as its equilibrium strategy (separating vs. pooling). In addition, we find that improving the accuracy of the deployed signaling mechanism will always benefit the responsible firm but may or may not hurt the irresponsible firm. Our results suggest that addressing the information asymmetry issue is the key to align consumers' goodwill with firms' responsible corporate behaviors. In particular, concerned parties should first exert efforts to create transparency in firms' sustainability practices before making investments to educate consumers and influence their purchasing behaviors. We further provide several model discussions to both confirm the robustness of our results and derive additional managerial insights.

Date 04.10.2019
Time 11:00 am - 12:15 pm
Venue Room 3003, LSK Business Building

Computer Vision and Metrics Learning for Hypothesis Testing: An Application of Q-Q Plot for Normality Test
Prof. Ke-Wei Huang, Department of Information Systems and Analytics, National University of Singapore

Date 20.09.2019
Time 10:30 am - 12:00 pm
Venue ISOM Conference Room 4047, LSK Business Building

Data-driven Consumer Debt Collection via Machine Learning
Dr Qingchen Wang, Innovation and Information Group, Faculty of Business and Economics, The University of Hong Kong

This paper develops and tests a data-driven framework for the scheduling of outbound calls made by debt collectors. We determine on a daily basis which debtors should be called to maximize the amount of debt recovered in the long term, under the constraint that only a limited number of phone calls can be made each day. Our approach is inspired by Markov decision processes, but given the intractability arising from having an extensive state space, we approximate the value function based on detailed historical data through the use of machine learning. Specifically, we predict the likelihood with which a debtor in a particular state is going to settle his debt and use this as a proxy for the value function. Based on this approximation, we compute for each debtor the marginal value of making a call, and prescribe phone calls by prioritizing debtors in states that have the highest marginal value. This approach is flexible, and is able to exploit all information available in the data regardless of complexity. We validate our methodology using a controlled field experiment conducted with 921 real debtors in partnership with a mid-sized debt collection agency. The results show that our data-driven policy substantially outperforms the incumbent calling policy that has been used in business practice for many years—collecting more debt while using substantially fewer resources. To conclude, the performance uplift from the experiment suggests that our framework is able to learn from the data and improve calling decisions.

Date 06.09.2019
Time 11:00 am - 12:15 pm
Venue Room 1001, LSK Business Building

Achieving High Individual Service-Levels without Safety Stock? Optimal Rationing Policy of Pooled Resources
Professor Jiawei Zhang, Professor of Technology, Operations, and Statistics and Robert Stansky Research Faculty Fellow, Stern School of Business, New York University

Resource pooling is a fundamental concept that has many applications in Operations Management for reducing and hedging uncertainty. An important problem in resource pooling is to decide (1) the capacity level of pooled resources in anticipation of random demand of multiple customers and (2) how the capacity should be allocated to fulfill customer demands after demand realization. In this paper, we present a general framework to study this two-stage problem when customers require individual and possibly different service-levels. Our modeling framework generalizes and unifies many existing models in the literature.

We propose a simple randomized rationing policy for any fixed feasible capacity level. Our main result is the optimality of this policy for very general service-level constraints, including Type-I and Type-II constraints and beyond. The result follows from a semi-infinite linear programming formulation of the problem and its dual. We also prove optimality of index policies for a large class of problems when the set of feasible fulfilled demands is a polymatroid.

Date 29.07.2019
Time 11:00 am - 12:15 pm
Venue Room 3003, LSK Business Building

A Primal-Dual Approach to Analyzing ATO Systems
Prof Yehua Wei, Fuqua School of Business, Duke University

We study assemble-to-order (ATO) problems from the literature. ATO problems with general structure and integrality constraints are well known to be difficult to solve, and we provide new insight into these issues by establishing worst-case approximation guarantees through various primal-dual analyses and LP rounding. First, we relax the one-period ATO problem using a natural newsvendor decomposition and use the dual solution for the relaxation to derive a lower bound on optimal cost, providing a tight approximation guarantee that grows with the maximum product size in the system. Then, we present LP rounding algorithms that achieve both asymptotic optimality as demand grows large, and a 1.8 approximation factor for any problem instance. Finally, we demonstrate that our one-period LP rounding results can be extended to analyze dynamic ATO problems. Specifically, we use our rounding scheme to develop an asymptotically optimal integral policy for dynamic ATO problems with backlogging and identical component lead-times.

Date 04.07.2019
Time 11:00 am - 12:15 pm
Venue Room 3003, LSK Business Building

The Overlooked Benefit of Consumer Showrooming for a Physical Retailer
Professor Lin Hao, Department of Information Systems and Operations Management, University of Washington

Date 27.06.2019
Time 10:30 am - 12:00 noon
Venue ISOM Conference Room 4047, LSK Business Building

Proactively Managing the Early Stages of an IS Academic Career
Professor Arun Rai, Regents’ Professor of the University System of Georgia, Robinson and Harkins Chairs of Information Systems, Robinson College of Business, Georgia State University

Date 25.06.2019
Time 10:30 am - 12:00 noon
Venue ISOM Conference Room 4047, LSK Business Building

To Brush or Not to Brush: Product Rankings, Customer Search, and Fake Orders
Prof Luyi Yang, Carey Business School, Johns Hopkins University

“Brushing”---the practice of online merchants placing fake orders of their own products to artificially inflate sales on e-commerce platforms---has recently received widespread public attention. On the one hand, brushing enables merchants to boost their rankings in search results, because products with higher sales volume are often ranked higher. On the other hand, rankings matter because search frictions faced by customers narrow their attention to only the few products that show up at the top. Thus, fake orders from brushing may affect customer choice. We build a stylized model to understand merchants’ strategic brushing behavior and its welfare implications. We consider two competing merchants selling substitutable products (one of high quality, the other of low quality) in an evolutionary sales-based ranking system that assigns a higher ranking to a product with higher sales. In principle, such an adaptive system improves customer welfare relative to a case in which products are randomly ranked, but it also triggers brushing as an unintended consequence. Since the high-quality merchant receives a favorable bias in the sales-based ranking, he mainly has a defensive brushing incentive, whereas the low-quality merchant mostly has an offensive brushing incentive. As a result, brushing is a double-edged sword for customers. It may lead customer welfare to be even lower than what it would be in a random-ranking system, but in some other cases, it can surprisingly improve customer welfare. If brushing is more difficult for merchants (e.g., due to tougher regulations), it may make customers worse off as it attenuates brushing by the high-quality merchant but induces the low-quality one to brush more aggressively. If search is easier for customers (e.g., due to improved search technologies), it can actually hurt them as it may disproportionately discourage the high-quality merchant from brushing.

Date 14.06.2019
Time 11:00 am - 12:15 pm
Venue Room 1003, LSK Business Building

Digital Multisided Platforms and Women's Health: An Empirical Analysis of Peer-to-Peer Lending and Abortion Rates
Professor Anandasivam Gopal, Dean's Professor of Information Systems, Robert H. Smith School of Business, University of Maryland

Date 10.06.2019
Time 10:30 am - 12:00 noon
Venue Room 1003, LSK Business Building

Operational Risk Management: A Stochastic Control Framework with Preventive and Corrective Controls
Professor Michael Pinedo, Julius Schlesinger Professor of Operations Management, Stern School of Business, New York University

We propose a general modeling framework for operational risk management of financial firms. We consider operational risk events as shocks to a financial firm's value process, and then study capital investments in preventive and corrective controls to mitigate risk losses. The optimal decisions are made in three scenarios: (i) preventive control only, (ii) corrective control only, and (iii) joint controls. We characterize the optimal control policies within a general modeling framework that comprises these three scenarios, and then discuss an exponential risk reduction function. We conclude our work with an application of our model to a data set from a commercial bank. We find that through a proper investment strategy, we can achieve a significant performance improvement, especially when the risk severity level is high. Moreover, with controls, the value of the firm tends to increase relative to the value of the firm without controls. Hence the controls are essentially smoothing out the jump losses and increasing the value of the firm. At the bank we analyze we find that with a joint control strategy the bank can achieve profit increases from 7.45\% to 11.62\% when the risk reduction efficiencies of the two controls are high. In general, our modeling framework, which combines a typical operational risk process with stochastic control, may suggest a new research direction in operations management and operational risk management.

(This paper is joint work with Yuqian Xu (University of Illinois - Urbana Champaign) and Lingjiong Zhu (Florida State University)

Date 04.06.2019
Time 11:00 am - 12:15 pm
Venue Room 1001, LSK Business Building

Optimal Monitoring Schedule in Dynamic Contracts
Professor Peng Sun, Fuqua School of Business, Duke University

Consider a setting in which a principal induces effort from an agent to reduce the arrival rate of a Poisson process of adverse events. The effort is costly to the agent, and unobservable to the principal, unless the principal is monitoring the agent. Monitoring ensures effort but is costly to the principal. The optimal contract involves monetary payments and monitoring sessions that depend on past arrival times. We formulate the problem as a stochastic optimal control model and solve the problem analytically. The optimal schedules of payment and monitoring demonstrate different structures depending on model parameters, and may involve monitoring for a random period of time.  Overall, the optimal dynamic contracts are simple to describe, easy to compute and implement, and intuitive to explain.

Here is a link to the paper: https://faculty.fuqua.duke.edu/~psun/bio/monitoring.pdf

Date 29.05.2019
Time 11:00 am - 12:15 pm
Venue G001, LSK Business Building

Platform Strategy and Operations in China: A Perspective Based on Economic Frictions
Professor Leon Zhu, Marshall School of Business, University of Southern California

Platforms are critical components of China’s economic system and often help regulate the system performance. In the first part of the talk, we consider a taxi-hailing platform that provides free information to taxi drivers. Upon receiving a rider's request, the platform broadcasts that rider's origin and destination to idle drivers, who accept or ignore the request depending on profitability considerations. We show that providing such information may reduce drivers' equilibrium profit. Hence information provision is a double-edged sword: drivers may strategically idle themselves to compete for the more profitable riders. We propose a dispatching policy under which, for large systems, broadcast information can achieve the first-best outcome. In the second part of the talk, we consider online platforms that enable service providers who offer customized services. The competition among the service providers may lead to a unique equilibrium under which all the service providers shirk. To address this opportunism problem, we propose a novel platform design where, from its inception, the platform deliberately limits the number of service providers below a certain threshold, even if they are homogenous.

Date 22.05.2019
Time 11:00 am - 12:15 pm
Venue Room 3001, LSK Business Building

Joint Statistics Seminar - Preferential Attachment and Neutral Random Graphs: Statistically Useful Generative Models of Network Data
Prof. Benjamin Bloem-Reddy, Department of Statistics, University of Oxford

Preferential attachment (PA) and other probabilistic generative models of network growth have been popular for their ability to explain large-scale phenomena from simple interaction mechanisms.  However, PA has been of limited use as a statistical model, due to its lack of exchangeability: in a statically observed network with n edges, inference requires considering all n! possible edge arrival orders.  Moreover, in models based on forms of exchangeability, inference algorithms benefit from an edge-decoupled representation, in which all dependence between edges is captured by some latent quantity; no such representation is known for PA models.  I will describe my work toward making PA useful as a statistical model: an edge-decoupled representation for a class of generalized PA models is established, and it reveals probabilistic structure, called left-neutrality, that can be exploited for efficient inference algorithms even in the presence of unknown edge arrival order.  Furthermore, the edge-decoupled representation endows the PA model with a set of interpretable model parameters.  Finally, I will describe how exchangeability still plays a role, despite PA's non-exchangeability.

This work was done in collaboration with Christian Borgs, Jennifer T. Chayes, Adam Foster, Emile Mathieu, Peter Orbanz, and Yee Whye Teh.

 

Date 17.05.2019
Time 11:00 am – 12:00 noon
Venue Room 4047 (LSK Business Building)

Local Inference in Additive Models with Decorrelated Local Linear Estimator
Dr. Zijian Guo, Department of Statistics, Rutgers University

Additive models, as a natural generalization of linear regression,  have played an important role in studying nonlinear relationships. Despite of much recent progress on additive models, the statistical inference problem in additive models has been much less understood. Motivated by inference for the exposure effect, we tackle the statistical inference problem for $f_1'(x_0)$ in additive models, where $f_1$ denotes the univariate function of interest and $f_1'(x_0)$ denotes its first order derivative evaluated at a specific point $x_0$. The main challenge for this local inference problem is due to the additional uncertainty of estimating other nuisance functions. To address this, we propose a decorrelated local linear estimator, which is particularly useful in reducing the effect of the estimation error related to the nuisance functions on the estimation accuracy of $f'_1(x_0)$. We establish the asymptotic limiting distribution for the proposed estimator and then construct confidence interval and conduct hypothesis testing for the estimand $f_1'(x_0)$. The variance level of the asymptotic limiting distribution is of the same order as that for the nonparametric regression while the bias of the proposed estimator is jointly determined by how well we can estimate the nuisance functions and  the relationship between the variable of interest and the nuisance variables. The method is developed for general additive models and is demonstrated in high-dimensional sparse additive model.

Date 14.05.2019
Time 11:00 am – 12:00 noon
Venue Room 4047 (LSK Business Building)

Who Gets the Attention? The Interactions among Similar Social Media Content
Professor Bin Gu, Earl and Gladys Davis Distinguished Professor, W P Carey School of Business, Arizona State University

Date 15.04.2019
Time 2:30 - 4:00 pm
Venue ISOM Conference Room 4047, LSK Business Building