UC Berkeley

Chapter one examines the effects of bundling social incentives with promotions. When

using these “social promotions,” a firm must decide whether or not to require customers to

share the promotion on social media with friends. On the one hand, this requirement may make

it easier for the firm to get the word out. On the other hand, this requirement may inconvenience

customers and lower their propensity to purchase. To conduct valid inference on this trade-off

between the costs and benefits, I devise novel empirical strategies that flexibly accommodate

the amount of information available to a firm. Through a field experiment on a Chinese online

grocery site, randomized both at the city and individual level, I find that social promotions

can serve as an effective growth strategy. The social obligation can benefit a firm through two

channels. First, the customers who do share add additional value through both new customer

acquisition and existing customer retention. Second, the firm does not lose profit from the

customers who choose not to purchase the promoted products when forced to share (but would

otherwise purchase the promoted products without mandatory sharing); they contribute instead

by purchasing other substitutes from the same site. Furthermore, different types of customers

respond to the social incentive differently and I present results on customer heterogeneity.

Chapter two expands on the targeting question in social promotions. This paper examines

how a firm should target a social incentive on top of a standard promotion, i.e. also requiring

the customer to share the promotion with friends on social media to qualify for the discounts.

A major challenge in this targeting question is that the targeting decision on a focal customer

also affects how many leads will be generated. Therefore, tackling this targeting question

requires a different framework compared with that of standard targeting. This paper compare

a few targeting schemes, including a post-experiment strategy based on an “effective” profit

idea and two contextual bandits that balance exploration and exploitation real time, namely

LinUCB and LinTS.We find that in our offline simulations, the two contextual bandits perform

similarly and better than post-experiment targeting. Targeting the social incentive to certain

type of customers can significantly increase firm profitability.

Chapter three discusses an empirical case study on bundle design and pricing. Unlike what

conventional bundle pricing theory suggests, we find that on the largest Chinese e-commerce

platform Taobao.com, a bundle tends to cost more than purchasing each bundle item from the

same online market separately. These bundle premiums are prevalent in multiple product categories,

such as digital cameras, iPads, cell-phones, and video game consoles. Conservatively

speaking, in the digital single-lens reflex cameras market, “premiumed bundles” account for

30% of the market share. Furthermore, the magnitude of the premium is also beyond that of rational

expectation. For example, for Canon 700D bundles, the average premium is 17 USD, or

4% the base good price. These bundle premiums can have significant implications for consumer

welfare and platform design. This paper systematically documents these empirical regularities.