Seungjin Whang’s research while affiliated with Stanford University and other places

This publication contains reprint articles for which IEEE does not hold copyright. Full text is not available on IEEE Xplore for these articles.

In the past 20 years the Operations Management and Information Systems (OM/IS) fields have widely applied game theory these fields had previously been more focused on analyzing and improving the performance of a physical system (e.g., queueing or inventory systems) from the decision theoretic perspective. This chapter gives an overview of the existing OM/IS literature applying game theory. We focus on five topics: (i) time-based competition, (ii) priority pricing for a queueing system, (iii)manufacturing/marketing incentives, (iv) incentives for information sharing in a supply chain, and (v) distortion and alignment in a horizontal supply chain. On each topic, we review one or two works at some length and list other related works as references.

In this paper we explore the underlying consumer heterogeneity in competitive markets for subscription-based IT services that exhibit network effects. Insights into consumer heterogeneity with respect to a given service are paramount in forecasting future subscriptions, understanding the impact of price and information dissemination on market penetration growth, and predicting adoption path for complementary products that target the same customers as the original service. Employing a continuous-time utility model, we capture the behavior of a continuum of consumers who are differentiated by their intrinsic valuations from using the service. We study service subscription patterns under both perfect and imperfect information dissemination. In each case, we first specify the conditions under which consumer rational behavior supported by the utility model can explain a general observed adoption path, and if so, we explicitly derive the analytical closed-form expression for the consumer valuation distribution. We further explore the impact of awareness and distribution skewness on adoption. In particular, we highlight the practical forecasting importance of understanding the information dissemination process in the market as observed past adoption may be explained by several distinct awareness and heterogeneity scenarios which may lead to divergent adoption paths in the future. Moreover, we show that in the later part of the service lifecycle the subscription decision for new customers can be driven predominantly by information dissemination instead of further price markdowns. We also extend our results to time-varying consumer valuation scenarios. Furthermore, based on our framework, we advance a set of heuristic methods to be applied to discrete-time real industry data for estimation and forecasting purposes. In an empirical exercise, we apply our methodology to the Japanese mobile voice services market and provide relevant managerial insights from the analysis.

Wireless telecommunications have become over time a ubiquitous tool that not only sustains our increasing need for flexibility and efficiency, but also provides new ways to access and experience both utilitarian and hedonic information goods and services. This paper explores the parallel market evolution of the two main categories of wireless services - voice and data - in leading technology markets, inspecting the differences and complex interactions between the associated adoption processes. We propose a model that addresses specific individual characteristics of these two services and the stand-alone/add-on relationship between them. In particular, we acknowledge the distinction between the nonoverlapping classes of basic consumers, who only subscribe to voice plans, and sophisticated consumers, who adopt both services. We also account for the fact that, unlike voice services, data services rapidly evolved over time due to factors such as interface improvement, gradual technological advances in data transmission speed and security, and the increase in volume and diversity of the content and services ported to mobile Internet. Moreover, we consider the time gap between the market introduction of these services and allow for different corresponding consumer learning curves. We test our model on the Japanese wireless market. The empirical analysis reveals several interesting results. In addition to an expected one-way effect of voice on data adoption at the market potential level, we do find two-way codiffusion effects at the speed of adoption level. We also observe that basic consumers impact the adoption of wireless voice services in a stronger way compared to sophisticated consumers. This, in turn, leads to a decreasing average marginal network effect of voice subscribers on the adoption of wireless voice services. Furthermore, we find that the willingness of voice consumers to consider adopting data services is positively related to both time and penetration of 3G-capable handsets among voice subscribers.

Postponement — delaying the point of product differentiation in a production process — has been found to be an effective strategy in face of increasing product variety. One value of postponement is flexibility. While the product is still in its generic form, the manufacturer can observe the latest demands for the different end-products and, at the postponement point, determine how much to customize the generic product for each end-product form. Hence, the value of postponement comes from the resolution of uncertainty of demands up to the postponement point. Another value of postponement is that the forecast accuracy for the demand of the final products in the distant future can also be improved. In this case, postponement enables the decision concerning the final customization of the end-products to be made at the time when the demand forecasts of the end-products are improved. We quantify both values of postponement in this paper, and discuss their relative magnitudes as product variety increases.

This paper studies the incentives behind the adoption of radio-frequency identification (RFID) in a supply chain. One prominent feature of RFID technology is that once RFID tags are attached on the items at an upstream site, the same tags can be reused at its downstream sites at lower or zero variable cost. This creates an interesting, one-sided "free-rider" problem, where the downstream would wait to free-ride on the upstream's first move, but not vice versa. Using a stylized game-theoretic model, we characterize the equilibrium strategies of the two firms. Compared to the first-best solution, firms in equilibrium tend to adopt too late. We then study the dual benefits of technology coordination between the two firms and find that it would not only save redundant costs of putting tags, but also speed up the downstream's RFID adoption. We also show that the equal-cost-split arrangement shifts the benefit of free-riding to the upstream, thereby mitigating the negative impacts in many cases. But it may distort the market when it operates in the optimal manner. The general message of the model is that technology coordination and cost-split each contribute to the mitigation of the free-rider problem in RFID adoption.

Suppliers in a variety of industries today face the challenge of managing their business where the utilization of their capacities fluctuates dramatically over time. The fluctuations can be attributed to the business cycle of the economy, as well as to the amplification of demand variability as one moves upstream (i.e., the bullwhip effect). In this chapter, we investigate the source of the fluctuations by analyzing a two-tier supply chain where the supplier serves many clients whose demands are subject to individual trends and the business cycle of the general economy. We present conditions under which the bullwhip effect or the stabilizing effect of the clients’ orders is felt by the supplier. We also analyze how the supplier can build an efficient client portfolio by analyzing the impact of a new client on the expected profit to the supplier in a newsvendor setting. We derive the key performance indicators that can guide a supplier in the right direction of the client portfolio. Thus, by understanding the clients’ ordering behaviors and its impact on capacity decisions, the supplier can strategically select an efficient client portfolio, so that the risk-neutral supplier would maximize the expected long-term profit.

This special issue of the Production and Operations Management journal offers a sample of ongoing research arising from the deployment of radiofrequency identification technology in organizational operations. The articles were selected to cover a spectrum of application areas and methodologies. There is also an attempt to identify interesting open areas for further investigation.