市场营销系学术讲座系列第一百零二讲

主 题：Image Portfolio and Demand in the Sharing Economy

主讲人: 何佳秀 (SUNY Buffalo State)

时 间：2019-01-09 14:00

地 点：明德商学楼1008室

语 言：中、英文

The rapid growth of peer-to-peer marketplaces has greatly facilitated the short-term rental of durable goods, creating the so-called the “sharing economy,” which represents a new business model built around the sharing of human, physical, and social resources. Sharing economy companies have reshaped their corresponding industries; for example, Airbnb, Inc., the context of our study, creates a global network through which anyone-anywhere could rent a living space to/from others for short periods. A key difference between sharing economy offerings, such as Airbnb, and traditional firms, such as chain hotels, is the choice between standardization and an individualized experience. Chain hotels offer standardized rooms and services across markets with standardized marketing efforts. Conversely, a key selling point of Airbnb is the unique and individualized experience that each listing provides. This uniqueness creates superior value for many travelers, but at a cost; without standardization, there is greater uncertainty regarding the overall quality of the property. Therefore, travelers form their expectations mostly based on information provided by each property’s host. Yet, there is substantial variation in the way Airbnb hosts display their properties on the platform visually and in text. For consumers (i.e., travelers), there is much less information to rely on from Airbnb than from hotels—fewer comments, fewer ratings, and no brand names to trust. As a result, Airbnb users tend to rely heavily on property images to make decisions. Indeed, Airbnb constantly highlights the importance of these images, noting, “high-quality photos are a great way to showcase your space, set guests expectations, and increase bookings.” In this research, we investigate the effects of visual information on demand for properties listed on Airbnb. The task of mining information from images has long been challenging because of the difficulty of extracting variables from visual data. Recent advances in machine learning techniques, however, now enable us to extract several important measures from Airbnb photos using a computational aesthetics approach and convolutional neural networks (CNNs). We seek to estimate the causal effect of image components on property demand and to document the size of this effect. In the process, we seek to make several novel contributions to the relevant literature. First, recent Airbnb studies mostly focus on guest reviews, aesthetic measures of photos, and other property and host characteristics; we contribute to this literature by studying image portfolio and contents of images. We find that visual information has direct and practical implications for hosts’ operations. Second, we construct an algorithm that distinguishes blocked dates against booked dates; this allows a more accurate measurement of the true booking rate. Third, from a methodological perspective, we implement a unique identification strategy, where we match focal city with three other cities and obtain instruments from these “other cities” for each focal listing.