Research

My substantial research area focuses on digital marketing. My recent research projects inspect:

  • the sales lift of live commerce
  • effectiveness of live-streamers’ communication strategies
  • intrinsic vs. extrinsic motivations of content creators on social media

In terms of methodologies, I intertwine econometric methods with advances in computer science. I approach research questions using a variety of techniques, such as instrumental methods, maximum likelihood methods, Bayesian statistics, synthetic controls, natural language processing, and computer vision.

My long-term research goal is to seek innovative solutions to marketing challenges in the digital era. Specifically, I’d like to decode the vast amount of unstructured data that consumers encounter every single day, and investigate their impact on consumer behavior.

Working Papers:

  • A Deep Dive into Live-Streaming Sales Pitch. Zining Wang, Yanwen Wang, Shuai Yang, and Hongju Liu. Job Market Paper
    • This research aims to unravel the mystique of live shopping. While live-streaming technology has reshaped the retail industry, the dynamics within live channels remain largely unobserved, presenting a challenge in accurately evaluating the performance of live-streamers. To bridge this knowledge gap, we leverage a unique live-streaming dataset to characterize and investigate the sales impact of communication strategies employed by live-streamers. Specifically, we focus on three types of sales pitches: product, promotion, and customer relationship management (CRM). We develop an automated, deep-learning-based framework to extract granular textual, visual, and vocal features from 3,057 live product- video pairs broadcasted by 111 live-streamers on a leading online marketplace in China in 2020. Our findings indicate that the usage of product sales pitches significantly impacts live-streaming sales, such that on average, a 10% increase in the ratio of product-related content to CRM-related content triggers an 11.0% increase in live-streaming sales. This sales lift of product-related content, however, wanes as the follower size of live-streamers decreases, and can even be negative for micro live-streamers. In general, live-streamers are unlikely to achieve higher sales by giving more priority to promotion sales pitches. Our research takes the initiative to evaluate the communication strategies utilized in the context of live commerce.
  • Sales Impacts of Live Streaming. Yanwen Wang, Shuai Yang, Hongju Liu, and Zining Wang. Reject and Resubmit at Marketing Science: Frontiers
    • This research investigates the sales impacts of live-streaming videos. We estimate live-streaming elasticities for 304 apparel brands by merging the brands’ product-level national sales with a complete record of product-level live-streaming activities from a leading online marketplace in China. Our empirical analysis suggests that live-streaming videos on e-commerce platforms have a positive sales impact on 64% of brands. Specifically, a 100% increase in the median brand’s live-streaming events will increase the sales of its promoted products by 16.8%. Meanwhile, live-streaming activity has no sales impact on 19% of brands and a negative sales impact on the remaining 17% of brands. Brand strength moderates the live-streaming effects. Strong brands are found to have the most responsive live-streaming elasticity. The effectiveness of live-streaming videos also varies across host types, with the live-streaming videos hosted by brands being the most effective, followed by star, micro, and platform live-streamers. The estimated live-streaming elasticities are robust given alternative model specifications and measures. The documented live-streaming elasticities support the emergence of live-streaming videos as a viable new form of communication channel for sales.
  • Toward a Thriving Two-Sided Ecosystem: The Ripple Effect of User Feedback on Content Creation. Zining Wang and Yanwen Wang. Finalist in the Inaugural 2020 RMP Data-Driven Research Challenge
    • The last two decades have witnessed a surge in content-sharing platforms, such as YouTube, Twitter, and TikTok. As the demand for user generated content (UGC) increases on a large scale, content-sharing platforms need to address a crucial problem: how to incentivize content creation? This work is aimed to investigate whether content creators have extrinsic motivations, and if so, to what extent could content creators be extrinsically versus intrinsically motivated. We use a latent class discrete choice model to analyze data from a content-sharing platform named Cloud Village. The findings show that extrinsic motivations play a crucial role in content generation. In particular, a higher level of user feedback could nudge creators to contribute more, whereas the effect of user feedback is stronger on prolific and inactive creators than on moderately prolific creators. In addition, content creators form an “engagement pyramid”: 1.8% of content creators are dedicated to content generation, 13.9% are moderately prolific, and 84.3% are rather inactive. Simulation studies suggest that given an increase in user feedback, the large group of inactive content creators could make a major contribution. The work sheds light on how to strike a balance between content creation and content consumption. The empirical findings reveal the potential of utilizing multi-objective recommender systems to create positive synergies between content consumers and content creators.
  • Insufficient Sleep and Price Sensitivity. Yitian Liang, Zining Wang, Zhongqiang Huang, Gerald J. Gorn, and Charles B. Weinberg. Available at SSRN.
    • Insufficient sleep is a widespread and chronic problem with well-documented con- sequences for health, economics, and decision making. Does lack of sleep also affect consumption behavior? In this paper, we introduce insufficient sleep as an antecedent of price sensitivity, a fundamental aspect of consumption that rarely has been linked to physiologically related factors. Using an 11-year large representative sample of U.S. households and exploiting sunset time as an exogenous proxy for sleep, we uncover a novel effect: lack of sleep increases price sensitivity over a wide range of grocery categories. A subsequent study that directly measures sleepiness further confirms this result. Across categories, the effect is greater for those with higher price sensitivity; within a category, the effect is greater for brands with higher relative prices but lower for brands with higher market shares. Our key finding that insufficient sleep is associated with increased price sensitivity is somewhat non-intuitive: managers surveyed a priori display divergent views on the issue. Our results shed light on the importance of physiologically related factors in consumption behavior and provide useful managerial implications.