SHUNYUAN ZHANG. Personal Website:

Transcription

1 SHUNYUAN ZHANG September 2018 Tepper School of Business Office: Posner 240B 5000 Forbes Avenue Cell: (765) Pittsburgh, PA Personal Website: EDUCATION Carnegie Mellon University, Tepper School of Business Pittsburgh, PA Ph.D in Marketing/Business Technology 2019 (Expected) Minor in Machine Learning Minor in Economics Purdue University West Lafayette, IN Ph.D in Physics 2014 University of Science and Technology of China Anhui, China Bachelor of Science 2008 RESEARCH INTEREST Topics: Sharing Economy, Crowdsourcing, Social Media Quantitative Marketing, Digital Marketing Methodologies: Dynamic Structural Modeling, Statistcial Modeling, Bayesian Inference Deep Learning, Computer Vision, Natural Language Processing JOB MARKET PAPER "Can Lower-Quality Images Lead to Greater Demand on Airbnb?",with Nitin Mehta, Param Vir Singh, Kannan Srinivasan, Dissertation Committee: Kannan Srinivasan (Chair), Param Vir Singh (Chair), Nitin Mehta, Tridas Mukhopadhyay, Anindya Ghose PUBLICATIONS Shunyuan Zhang, Param Vir Singh, Anindya Ghose, 2016, "A Structural Analysis of the Role of Superstars in Crowdsourcing Contests," Forthcoming at Information Systems Research. 1

2 September 2018 WORKING PAPERS Shunyuan Zhang, Nitin Mehta, Param Vir Singh, Kannan Srinivasan, 2018, "Can Lower-Quality Images Lead to Greater Demand on Airbnb?", Drafting Shunyuan Zhang, Dokyun Lee, Param Vir Singh, Kannan Srinivasan, 2017, "How Much Is an Image Worth? Airbnb Property Demand Estimation Leveraging Large Scale Image Analytics, Under 2nd round of review at Management Science. o Winner, Best Student Paper Award at CIST 2016 o Adobe Data Science Research Awards 2017 Shunyuan Zhang, Dokyun Lee, Param Vir Singh, Tridas Mukhopadhyay, 2018, Demand Interactions in Sharing Economies: Evidence from a Natural Experiment Involving Airbnb and Uber/Lyft, Revising for 2nd round of review at Management Science. WORK IN PROGRESS Shunyuan Zhang, Towards the Reasoning of Ads via Image Captioning: A Visual Question Answering (VQA) Approach, Shunyuan Zhang, "Who Benefits from Artificial Intelligence? An Empirical Analysis of Returns to Smart Pricing Algorithm on Airbnb, with Nitin Mehta, Param Vir Singh, Kannan Srinivasan, 2018 CONFERENCE/SEMINAR PRESENTATIONS INFORMS Marketing Science Conference, Philadelphia, PA, 2018 Purdue University Research Center For Open Digital Innovation, West Lafayette, IN, 2018 INFORMS Marketing Science Conference, Los Angeles, CA, 2017 Workshop on Information Systems and Economics (WISE), Seoul, Korea, 2017 INFORMS Annual Meeting, Houston, TX, 2017 Conference on Information Systems and Technology (CIST), Houston, TX, 2017 International Conference in Information Systems (ICIS), Dublin, Ireland, 2016 Workshop on Information Systems and Economics (WISE), Dublin, Ireland, 2016 Conference on Information Systems and Technology (CIST), Houston, TX, 2016 INFORMS Annual Meeting, Nashville, TN, 2016 INFORMS Annual Meeting, Philadelphia, PA, 2015 Conference on Information Systems and Technology (CIST), Philadelphia, PA, 2015 CONFERENCE/PROCEEDINGS PUBLICATIONS Shunyuan Zhang, Param Vir Singh, Anindya Ghose, 2015, "Analyzing the Role of Superstars in Crowdsourcing Contests: A Structural Model," Conference on Information Systems and Technology (CIST), Philadelphia, PA. 2

3 September 2018 Shunyuan Zhang, Dokyun Lee, Param Vir Singh, Kannan Srinivasan, 2016, How Much Is An Image Worth? An Empirical Analysis of Property s Image Aesthetic Quality on Demand at AirBNB, International Conference in Information Systems (ICIS), Dublin, Ireland. Shunyuan Zhang, Dokyun Lee, Param Vir Singh, Kannan Srinivasan, 2016, Professional versus Amateur Images: Investigating Differential Impact on Airbnb Property Demand, Conference on Information Systems and Technology (CIST), Nashville, TN. [Winner of Best Student Paper Award at CIST 2016] Shunyuan Zhang, Dokyun Lee, Param Vir Singh, Kannan Srinivasan, 2016, "Image Feature Extraction and Demand Estimation on Airbnb: A Deep Learning Approach", Workshop on Information Systems and Economics (WISE), Dublin, Ireland. Shunyuan Zhang,Dokyun Lee, Param Vir Singh, Tridas Mukhopadhyay, 2017,"The Sharing Effects of Sharing Economy: Evidence from a Natural Experiment Involving Airbnb and Uber/Lyft," Conference on Information Systems and Technology (CIST), Houston, TX. Shunyuan Zhang,Dokyun Lee, Param Vir Singh, Tridas Mukhopadhyay, 2017,"Analyzing the Demand Interaction in Sharing Economy", Workshop on Information Systems and Economics (WISE), Seoul, Korea. GRANTS, HONORS AND AWARDS ISMS Doctoral Consortium Fellow, 2018 Dipankar and Sharmila Chakravarti Doctoral Award, 2018 Adobe Data Science Research Awards, 2017 ISMS Doctoral Consortium Fellow, 2017 Best Student Paper Award, Conference on Information Systems and Technology, 2016 Dean s Research Fund, Tepper School of Business, 2016 William Larimer Mellon Fellowship, Carnegie Mellon University, SELECTED GRADUATE COURSEWORK Course Microeconomics I Microeconomics II Econometrics I Econometrics II Game Theory and Applications Analytical Models in Marketing Bayesian Statistics in Marketing Economics & Marketing Course Instructor Isa Hahalir Stephen Spear Joachim R. Groeger Karam Kang Isa Hahalir Kaifu Zhang Alan Montgomery 3

4 Analytical and Structural Models in Marketing Dynamic Structural Models of Marketing and Economics Business Technology Seminar on Estimating Structural Models Economics of New Emerging Technologies Seminar on Social Networks Methods and algorithms for determining relevance & trust in networked data Machine Learning Machine Learning Natural Language Processing Computer Vision (audit) Advanced Multimodal Machine Learning Deep Reinforcement Learning and Control (audit) September 2018 Kannan Srinivasan Guofang Huang Param Vir Singh Tridas Mukhopadhyay Param Vir Singh Wolfgang Gatterbauer Maria-Florina Balcan Tom Mitchell Alan W. Black David Mortensen Deva Ramanan Louis-Philippe Morency Ruslan Salakhutdinov TEACHING EXPERIENCE Instructor Pricing Strategy (Undergraduate), Tepper School of Business, 2018 Teaching Assistant Pricing Strategy (Undergraduate), Tepper School of Business, by Kaifu Zhang, 2016 Pricing Strategy (MBA), Tepper School of Business, by Kaifu Zhang, 2016 Digital Marketing and Social Media Strategy (MBA), Tepper School of Business, by Param Vir Singh, 2017 Topics in Deep Learning (Graduate), School of Computer Science (Machine Learning Department), by Ruslan Salakhutdinov, 2017 Modern Data Management (MBA), Tepper School of Business, by Tridas Mukhopadhyay, 2018 Strategic Information Technology (MBA), Tepper School of Business, by Tridas Mukhopadhyay, 2014 SKILLS Matlab, Python, SQL, PHP, R, STATA AD-HOC REVIEWER Information System Research, Conference on Information Systems and Technology (CIST), International Conference in Information Systems (ICIS) 4

5 September 2018 SELECTED RESEARCH ABSTRACTS Can Lower-Quality Images Lead to Greater Demand on Airbnb? Joint with Nitin Mehta, Param Vir Singh, and Kannan Srinivasan (Job Market Paper) Abstract: We investigate how AirBnB hosts make decisions on the quality of property images to post. Prior literature has shown that the images play the role of advertisements and the quality of the images have a strong impact on the present demand of the property as compared to lower quality amateur images, high quality professional images can increase the present demand by 14.3% on matched samples (Zhang et al. 2018). However, the reality is that there exist a large number (approximately two-thirds) of amateur (low-quality) images on Airbnb. One possible explanation is that these images are costly for the hosts, as most of them are amateur photographers. However, this does not completely explain the result in 2011, AirBnB started offering highest-quality professional images for free to all the hosts by sending their professional photographers to the property and shoot, process and post the photos for the hosts. To AirBnB s surprise, only 30% of the hosts used the AirBnB professional photography program. We posit that the host s decision on what quality of images to post depends not only on the advertising impact of images on the present demand and on the cost of images, but also on the impact of images on the future demand. Thus, some hosts would be hesitant to post high-quality images because they can create unrealistically high expectations for the guests, especially if the actual property is not as good as what the images portray and if the hosts are unable to provide a high level of service to match those expectations. This would result in the satisfaction level of guests to decrease, who would then leave a bad review or not write any review at all; and since the number/quality of reviews is one of the key drivers in generating new bookings, this will adversely affect the future demand. In this paper, we attempt to disentangle the aforementioned factors that influence the host s decision on the type of photographs to post, and explore policies that AirBnB can employ to improve the hosts adoption of professional photos and thereby improve the profitability of both the hosts and AirBnB. To do so, we build a structural model of demand and supply, where the demand side entails modeling of guests decisions on which property to stay, and the supply side entails modeling of hosts decisions on what quality of images to post and what level of service to provide in each period. We estimate our model on a unique one-year panel data consisting of 2,421 Airbnb properties in New York where we observe hosts monthly choices of the quality of images posted and the service that they provided. Our key findings are: First, guests who pay more attention to images tend to care more about reviews, revealing an interesting trade-off problem for the hosts. Second, hosts incur considerable costs for posting above-average quality of images. Third, hosts are heterogenous in their abilities in investing service effort. In counterfactual analyses we simulate Airbnb properties assuming they all start with entry state and low-level images. We then compare the impact of the current policy (offering free high-level images to hosts) and of a proposed policy (offering free medium-level images to hosts) on the average property demand. We show that the proposed policy, though dominated by the current policy in the short-run (for the first four periods), outperformed the currently policy in the long-run (12.9 % vs 6.0%). The interpretation is that, medium-level images, compared to highlevel images, despite forming a smaller expected utility for the consumers, has a greater effect on property demand in the long-run as they, with lower risks of creating a dissatisfactory gap, help hosts to obtain new reviews. Moreover, individual hosts who might end up using amateur (low-level) images to avoid the dissatisfactory gap under the current policy, now use 5

6 September 2018 free medium-level images to make more revenues under the proposed policy. In the second counterfactual, we explore an alternative policy in which AirBnb were to offer a menu of image quality choices for free. The menu includes both highand medium- level of property images (images examples are provided) and allow the hosts to self-select which program they want. Comparing with the proposed policy in the first simulation, we find that this policy performance the best in the long-run by improving average property demand by 14.0%. How Much is an Image Worth? Airbnb Property Demand Analytics Leveraging A Scalable Image Classification Algorithm Joint with Dokyun Lee, Param Singh, and Kannan Srinivasan (Under 2 nd round of review, Management Science) Abstract: We investigate the economic impact of images and lower-level image factors that influence property demand at Airbnb. Using Difference-in-Difference analyses on a 16-month Airbnb panel dataset spanning 7,711 properties, we find that units with verified photos (taken by Airbnb s photographers) generate additional revenue of $2,521 per year on average. For an average Airbnb property (booked for % of the days per month), this corresponds to 17.51% increase in demand due to verified photos. Leveraging computer vision techniques to classify the image quality of more than 510,000 photos, we show that 58.83% of this effect comes from the high image quality of verified photos. Next, we identify 12 interpretable image attributes from photography and marketing literature relevant for real estate photography that capture image quality as well as consumer taste. We quantify (using computer vision algorithms) and characterize unit images to evaluate the economic impact of these human-interpretable attributes. The results reveal that verified images not only differ significantly from low-quality unverified photos, but also from high-quality unverified photos on most of these features. The treatment effect of verified photos becomes insignificant once we control for these 12 attributes, indicating that Airbnb s photographers not only improve the quality of the image but also align it with the taste of potential consumers. This study suggests there is significant value in optimizing images in e-commerce settings on these attributes. From an academic standpoint, we provide one of the first large-scale empirical evidence that directly connects systematic lowerlevel and interpretable image attributes to demand. This contributes to, and bridges, the photography and marketing (e.g., staging) literature, which has traditionally ignored the demand side (photography) or did not implement systematic characterization of images (marketing). Lastly, these results provide immediate insights for housing and lodging e- commerce managers (of Airbnb, hotels, realtors, etc.) to optimize product images for increased demand. Demand Interactions in Sharing Economies: Evidence from a Natural Experiment Involving Airbnb and Uber/Lyft Joint with Dokyun Lee, Param Singh, and Tridas Mukhopadhyay (Major Revision, Management Science) Abstract: The existing research has largely focused on the impact of sharing economy on incumbent industries while ignoring the interactions among sharing economies. In this study, we examine how ride sharing services such as Uber and Lyft affect the demand for home sharing services such as Airbnb. Our identification strategy hinges on a natural experiment where Uber and Lyft exited Austin in May 2016 in response to the introduction of new regulations in Austin that targeted ride sharing services. Applying the Difference-in-Difference approach on a 9-month balanced longitudinal data spanning 6

7 September ,300 Airbnb properties across 7 US cities, we find that the exit of Uber/Lyft led to a decrease of 9.6% in the Airbnb property demand, which is equivalent to a decrease of $6,482 in the annual revenue to the host of an average property. We further find that the exit of Uber/Lyft reduced the (geographic) demand dispersion of Airbnb. The demand became more concentrated in areas with access to better public transportation services. Moreover, the properties farther from downtown experienced greater decreases in their demand in the absence of Uber/Lyft. The results indicate that Uber and Lyft affect the demand for Airbnb properties primarily by reducing the transportation costs to and from Airbnb properties that otherwise have poor access to transportation services. A Structural Analysis of the Role of Superstars in Crowdsourcing Contests Joint with Param Singh and Anindya Ghose (Forthcoming, Information Systems Research) Abstract: While online crowdsourcing contests became quite popular only recently, they bear close resemblance to classical tournaments. Economic literature on tournament-style contests has highlighted a superstar effect where participants may exert less effort and perform worse in the presence of a superstar contestant (who consistently performs far superior to others), as their chances of winning are very low. While an individual would find it difficult to avoid superstars in tournaments, they can do so in crowdsourcing contests. Recent Research has documented that individuals avoid competing with superstars due to the adverse superstar effect on Topcoder.com. The presence of superstars often leads to thinly participated contests, which has led to concerns expressed by practitioners because researchers have shown that the chances of achieving a high-quality best solution for an innovation-related task increase with the number of participants in a contest. In this study, we argue that previous research on the role of superstars has considered superstars only as competitors and has largely ignored their role as a potential source to learn from. In crowdsourcing contests, the learning opportunities that participants provide to each other are particularly salient. For example, at Topcoder.com, participants interact with each other throughout the contest and can observe the best in action. Furthermore, participants who have submitted a solution gain access to the winning solution, which, along with the forum discussion, can help them learn what the winner did that bested them. Using a unique 50-month longitudinal panel data set on 1677 software design crowdsourcing contests, we illustrate a learning effect where participants are able to improve their skills (learn) more when competing against a superstar than otherwise. We show that an individual s probability of winning in subsequent contests increases significantly after she has participated in a contest with a superstar coder than otherwise. We build a dynamic structural model with individual heterogeneity where individuals choose contests to participate in and where learning in a contest happens through an information theory-based Bayesian learning framework. Counterfactual analysis suggests that instead of avoiding superstars, individuals should be encouraged to participate in contests with superstars early on, as it can significantly push them up the learning curve, leading to higher quality and a higher number of submissions per contest. Overall, our study shows that individuals who are willing to forego short-term monetary rewards by participating in contests with superstars have much to gain in the long term. 7

8 REFERENCES Kannan Srinivasan (co-chair) H.J. Heinz II Professor of Management, Marketing and Business Technologies Carnegie Mellon University Tepper School of Business Param Vir Singh (co-chair) Carnegie Bosch Associate Professor of Business Carnegie Mellon University Tepper School of Business Nitin Mehta Professor of Marketing The University of Toronto Rotman School of Management Anindya Ghose Heinz Riehl Chair Professor of Business The New York University Leonard N. Stern School of Business September