The Effects of Online Review Platforms on Restaurant Revenue, Consumer Learning and Welfare (2021) (PDF) Management Science (Marketing Area), 68 (11), P7793-8514
Cited by the Federal Trade Commission (FTC) in “Trade Regulation Rule on the Use of Consumer Reviews and Testimonials”; Insights at UBC Sauder Blog: https://www.sauder.ubc.ca/news/insights/online-reviews-can-make-or-break-independent-restaurants-near-highways-study
Measuring Deterrence Motives in Dynamic Oligopoly Games (2024) with Nathan Yang (PDF) Management Science (Marketing Area), 70 (6), P3381-4165
Insights at UBC Sauder Blog: https://www.sauder.ubc.ca/news/insights/major-coffee-chains-often-move-wall-out-competitors-study
Cited by the White House in 2024 Economic Report of the President: https://www.whitehouse.gov/wp-content/uploads/2024/03/ERP-2024.pdf
The Downmarket Impact of New Multifamily Housing: Evidence from a Honolulu Condo Tower (2026) with Justin Tyndall and Emi Kim (PDF). Forthcoming at Real Estate Economics.
Featured in The Atlantic article "High-End Construction Really Does Help Everyone".
Working Papers
The Effect of Quality Disclosure on Firm Entry and Exit Dynamics: Evidence from Online Review Platforms (2026) with Ying Bao and Matthew Osborne (PDF). Minor Revision at Marketing Science.
Included in NBER Digital Economics & AI Spring Meeting 2024 and Summer Institute in Competition Strategy 2024
Abstract: This paper examines how quality disclosure reshapes market structure through firm entry and exit. Quality disclosure affects entry and exit through a direct effect (consumers learn a firm's quality) and a competition effect (consumers learn competitors' qualities and the overall distribution). Because market fundamentals determine which effect dominates, disclosure's equilibrium impact on entry, exit, and quality varies across markets. We study these mechanisms in the Texas restaurant industry by combining administrative data from 1995 to 2019 on the universe of full-service restaurants with online-review histories, exploiting staggered review-platform penetration. Using pre-treatment demand and entry-cost proxies, we classify markets into three types and document sharply different patterns. In medium-sized markets, disclosure generates moderate quality sorting for entrants without significantly impacting incumbent exit. In highway-exit markets, disclosure encourages entry and reduces incumbent exit across all quality levels. In metro markets, disclosure operates as a strong quality filter: low-quality entry falls while high-quality entry rises, and exit likelihoods decrease slightly. Consistent with these dynamics, metro markets experience an aggregate quality upgrade: a 100% increase in review activity raises the market-wide share of high-quality independent restaurants by roughly 7.3 percentage points. Overall, the effects of disclosure on market structure are strongly market-dependent.
Shifting Standards or Changing Experiences? Unraveling Review Polarization via LLMs (2026) with Chunhua Wu and Baohong Sun (PDF). Revise & Resubmit at Journal of Marketing Research.
Abstract: Online reviews are becoming increasingly polarized, as extreme 1-star and 5-star ratings crowd out moderate evaluations. This paper investigates whether this shift reflects changes in consumer experiences or a drift in rating standards. Using approximately four million Yelp restaurant reviews across 11 metropolitan markets, we estimate a Bayesian neural network model that maps review content---text and photos---into star ratings through time-varying thresholds. The model separates changes in what consumers express (the content effect) from changes in how those expressions are mapped to stars (the scale effect). We identify a previously unrecognized driver of polarization: standards drift, a market-wide recalibration of the rating scale that persists after accounting for reviewer and business composition. Standards drift explains approximately 77% of the growth in 5-star ratings, whereas the rise in 1-star ratings is primarily a content phenomenon driven by increasingly negative expressions among occasional reviewers. This narrowing of the rating scale generalizes to Google Maps and Amazon, suggesting a pervasive shift in online rating environments. Because drifting standards systematically penalize established businesses relative to newer entrants, these findings motivate the use of temporally recalibrated ratings or content-based AI scores to enhance informativeness and fairness.
A Tax-Shaped Retail Landscape (2025) with Feng Chi, Mengwei Lin, and Nathan Yang (PDF)
Abstract: We investigate the impact that seemingly uniform tax policies have on shaping the retail landscape. Using data about all retail establishments in the United States, we show that while retailers are more likely to open establishments in markets with favorable state tax policies, it is primarily the largest chains that are contributing towards this effect on entry. Motivated by these empirical realities, we develop an entry model where firms are subject to taxes that impact their sunk costs of entry. This model is used to demonstrate that taxes can potentially amplify market dominance, such that retailers with preexisting size-based advantages are disproportionately more responsive to the tax policies. Furthermore, we show that revenue-maximizing tax levels that the government could set have the potential to exacerbate the dominance of strong firms. Finally, we provide calibrated model analysis based on data about home improvement retail chain entry. This calibration exercise illustrates the economic magnitude associated with asymmetric retail entry responses to tax policy.
Abstract: We develop the theoretical basis for implementing non-monotonicity tests in dynamic retail entry contexts, as a means to obtain suggestive reduced-form evidence of deterrence motives. With this theoretical motivation, we then conduct two empirical case studies. Our first empirical case study focuses on competition between taco chains in Texas. In particular, we show that the test statistics for the incumbent's deterrence motives appear to be different before and after the threat of a major competitor's arrival. For our second empirical case study, we explore potential heterogeneity of the deterrence motives in competition between hamburger chains in Canada across geographic markets with and without shopping centers. The inferred heterogeneity reveals that deterrence motives are noticeably muted in markets with shopping centers, which is consistent with the potential use of exclusionary clauses between shopping centers and incumbent firms.
It Pays to Be Lucky: Superstition, Product Labels, and Housing Markets with Mingduo Zhao.
When 3.5 Stars on Yelp Means the Best: Cultural Matching and Bias in Online Restaurant Reviews, with Nitin Mehta, Chunhua Wu, and Siying Wang.
The Effect of Ownership Structure on Service Response to Competition: Evidence from Restaurant Industry (2025) with Ying Bao, Jessica Gu, and Aric Rindfleisch