Essays on Pricing and Promotion Policies of Digital Platforms

Abstract

This dissertation investigates the impact of cross-network effects on optimal product/service segmentation pricing and dynamic promotion within digital platforms. Specifically, it examines (1) the implications of vertical segmentation for digital platforms in terms of profitability, sellers' profitability, and buyers' utility, and (2) the optimal dynamic platform promotion alongside user-generated promotion from the embryonic stage to maturity. These topics are addressed through game theoretic models, dynamic programming, and comparative statics to derive managerial insights. The thesis draws upon the literature of economics and marketing, particularly focusing on two-sided platforms, information asymmetry, and dynamic promotion. The dissertation comprises the following inter-related chapters: (1) Introduction, (2) Literature Review on Digital Platforms and Related Businesses, (3) Vertical Segmentation Implications for Digital Platforms, (4) Optimal Dynamic Platform Promotion Policy under Evolution, and (5) Conclusion. The introduction section discusses the importance of digital platforms in the modern economy. The literature review section examines the inception of digital platforms and the differences in marketing strategies, competition, product/service categorization, and business evolution compared to traditional businesses. Gaps in the literature that the thesis aims to address are identified. Chapter 3 considers two broad situations: when the platform sets the price or when the seller sets the price. Equilibrium outcomes for integrated or segmented markets are derived, along with outcomes under perfect or imperfect information about product/service quality. Chapter 4 employs dynamic programming to derive Euler equations linking optimal promotion across periods. Three cases are considered: buyers/sellers changing over time, buyers changing but sellers fixed, and a three-period game with buyers/sellers changing over time but platform promotion limited to the first two periods. MATLAB is used to code the dynamic programming model, followed by simulations to derive steady-state outcomes and conduct comparative statics. The conclusion chapter summarizes the two papers and identifies potential areas for future research.