Papers & Current Research

Scroll down to see more content

Published Papers

Abstract: In this study, the authors investigate the externalities of managers’ responses (MRs) to online reviews on popular travel websites. Specifically, the authors examine the effect of publicly responding to hotel guests’ reviews on subsequent reviewer ratings. The authors find that manager responses to negative reviews (MR-N) can significantly influence subsequent opinion in a positive way if those responses are observable at the time of reviewing. Notably, the findings show this externality to be negative for manager responses to positive reviews (MR-P). The authors conduct a topic analysis on review texts and corresponding MRs to study the moderating role of response tailoring on the opinion externalities of MR. The authors show that tailored MR amplifies the positive (negative) impact of MR-N (MR-P) on subsequent opinion. Intuitively, tailoring an MR-N adds specificity to the hotel’s complaint management strategy, bolstering the positive effects of MR-N on subsequent opinion. However, by highlighting specific positive elements of a review, managers’ intent for responding is brought into question as they take advantage of reviewers’ positive feedback to promote their hotel.

Working Papers

Abstract: In this study, we use a game theoretic model to argue that the presence of online reviews can lead to product quality improvements for independent firms selling experience goods. Exploiting heterogeneous review platform penetration across markets, we test the predictions of our model using a dataset covering 40 thousand U.S. hotels and show that markets with greater TripAdvisor penetration exhibit greater gains in independent hotel quality. Independent hotels located in median peak penetration TripAdvisor markets improved their quality by an average of .129 stars as measured using composite online travel agent (OTA) star ratings, erasing 41% of the advantage held by chains in the absence of online reviews. We address measurement noise challenges for quality and platform penetration using state space models to reveal persistent quality and platform penetration trends. Additionally, we resolve endogeneity due to potential unobserved confounds correlated with penetration and quality across markets and time. We do so by exploiting review platforms’ imperfect market definitions that divide areas of hotel agglomeration into separate review platform markets, thus quasi-exogenously assigning hotels in the same area to varying levels of online review exposure. Our research suggests that online reviews play an important role in facilitating competition on quality.

Abstract: Using a unique dataset from a large food sharing platform, we investigate the effect of supply side assortment size on users’ purchase probabilities. We find that users’ purchase probabilities are increasing in assortment size but at a decreasing rate. The initial increase in purchase probability is large. Users exposed to the mean 15 pages (150 options) of search results have an 8.4% higher purchase probability than those who only have access to one page of options. However, this effect plateaus rapidly with a peak lift in purchase probability compared to average session of 2.86% at 29 pages of options. Furthermore, choice overload exists in our empirical support, leading to a slight decrease in purchase probability versus the optimal assortment size of .1% at 32 pages. Surprisingly, the diminishing returns to assortment size are shown to be due to search costs rather than evaluation costs or decreasing marginal assortment diversification. Moreover, we show that users’ exposure to nearby offline options can offset the diminishing impact of assortment size on purchasing. To obtain these results, we resolve endogeneity challenges by introducing a variant of the border identification strategy that exploits discrete delivery distances featured in the data. We contribute to the literature by demonstrating the limits of a platform’s aggressive supply side growth, showing that search costs can dampen the purchase probability gains driven by assortment size leading to choice overload, and suggesting that users’ outside options can alleviate the concave effects of assortment size on purchase probabilities.

Abstract: The rise of intelligent conversation agents, or chatbots, are responsible for the dramatic decrease in remote customer service agent jobs. However, chatbots in their current form, are far from infallible. We theorize that there is an inherent trade-off between a chatbot’s response relevance and conversational efficiency in the standard knowledge-bank architecture. Knowledge bank size increases the relevance of successfully queried results, but also increases the difficulty of disambiguating user intents. This inherent trade-off leads to the development of unintelligent fail-safe artifacts such as user confirmations. We argue that, in order to improve user experience and satisfaction, we must decouple knowledge bank size from conversational efficiency. To achieve this, we design a new artifact that we dub sequential FAQ (sFAQ). sFAQ uses machine learning techniques to first discover common user service journey patterns, then leverage these learned patterns to predict likely subsequent inputs given any focal sequence of inputs. We show that by proactively suggesting potential questions to the user, we can reduce the need for natural language input and thus reduce the need to disambiguate user intent. We then use a novel application of regression discontinuity design (RDD) to study the causal impact of the eliminated reconfirmation dialogues on user satisfaction. Combined, we are able to demonstrate that by eliminating the unintelligent fail-safe artifacts such as user confirmations, sFAQ will cause user satisfaction increases. Our approach of combining predictive machine learning and casual econometric analysis enables us to open the black box for the underlying causal mechanism linking sFAQ and user satisfaction. This kind of mechanism identification would not be possible even with experimental testing in the field. Our methods and results have useful implications for chatbot applications and user interface design science.

Abstract: We apply causal inference methods from econometrics to study the entire causal chain from local governments’ shelter in place policies to constituents’ social distancing behaviors to the transmission of the novel coronavirus disease 2019 (COVID-19) in the United States. Without studying the mechanism chain, research that links SIP policies to public health outcomes may misattribute the effectiveness of SIP policies due to unobservable confounds unrelated to the behavioral changes caused by the policy. Contrary to conventional wisdom, we find that shelter in place policies only caused modest changes in the intended social distancing behaviors. State and county level shelter in place policies increased the median percent of time spent at home by 2.5%. In contrast, non-policy factors led to an increase of 5.14%. Furthermore, we show that crowdedness measured using detailed location tracking data, rather than time spent at home, explains infection growth rates. Encouragingly, shelter in place policies explain half the decrease (9.9%) in crowdedness, leading to a 1.5% decrease in infection growth rates. In light of our research, policy makers should consider the role that non-policy factors play in corralling the transmission of COVID-19 while concentrating resources to deter crowding behavior.

  • (Sub)optimality of managerial dynamic pricing (currently revising)

Abstract: This study contributes to the largely theoretical field of revenue management with an empirical investigation into the sub-optimality of managerial dynamic pricing policies as evidenced in the Las Vegas hotel market. We demonstrate that managers consistently choose prices that yield revenues approximately 25% below optimal levels. Specifically, managers appear to choose prices in a manner consistent with maximizing a mix of occupancy and revenue. We find support for the hypothesis that the unobservability of counterfactual revenues may drive managers’ suboptimal pricing policies when the hotel is expected to fill capacity. Additionally, we explore a novel managerial use of online reviews in pricing decisions and the effect of competitors’ pricing strategies on a focal hotel’s optimal prices. We discover that predicting mean reverting tendencies of online reviews can marginally improve the focal hotel’s bottom line during slow seasons. Similarly, we show that there is an economically significant impact of predicting competitors’ prices on the focal hotel’s pricing policies.

Ongoing Projects

I see my projects as centered around interesting datasets. Currently, I am interested in the following unique data:

  1. Online reviews – millions of hotel, restaurant, and other travel reviews collected with Alex Chaudhry from across multiple review platforms.
  2. Movies – millions of reviews, movie revenues across the globe at daily/weekly levels, movie scripts, trailer videos
  3. Survey opinions – YouGov brand tracking surveys.
  4. Networks – Currently analyzing an expansive dataset of partners, suppliers, and competitors relationships in the economy.
  5. Advertising data – Ad$pender data augmented by USPTO trademarks data.

Ongoing projects in order of progress:

  • The impact of movie trailers on box office performance. – with Alex Chaudhry
  • Brand tracking: online reviews versus surveys, divergent insights on brand performance. -with Alexander Chaudhry and Rex Du
  • Does online word of mouth displace offline word of mouth? Evidence from the U.S. movie industry. – with Alex Chaudhry and Seethu Seetharaman
  • Advertising, mergers, and the network topology of economic supply chains.
Advertisement