Publications

We study whether political polarization shapes US cities' disclosures on climate mitigation (i.e., net zero transition initiatives). Using textual analysis of cities' budgets and annual reports, we categorize these disclosures into two types: (i) tangible descriptions of ongoing projects and (ii) intent, the non-committal discussions of the high-level project purposes and future plans. We find that while there is partisan difference in the ongoing project disclosures, it remains stable over time, suggesting that political polarization doesn't affect the implementation of underlying projects. However, we observe a growing divergence in intent, which is inherently more discretionary as it reflects promises rather than concrete actions. Consistent with cities acting in accordance with the preferences of their stakeholders, we find similar results when we examine the cities' communications with the citizenry and disclosures directed at bondholders.

Problem definition: Consumers often enjoy displaying luxury consumption to signal their private wealth status. The emergence of social media has fueled such desire for status signaling. Meanwhile, the rising of e-commerce has made it easy for consumers to search and purchase cheap non-deceptive counterfeits to send a ``deceptive'' status signal, posing a new challenge to the luxury (status product) industry. Methodology/results: Motivated by these industry dynamics, we consider a market entry deterrence game between an incumbent status product firm (the firm) and a non-deceptive counterfeiter (the counterfeiter) who attempts to enter the market. A salient feature of our model is that the market demand is endogenously determined by a consumer status signaling subgame. Our analysis yields four main insights. First, we show that without counterfeits, the firm is strictly better off from the heightened motive of consumer status signaling; however, such benefit would be neutralized by the potential counterfeiter entry. Second, we find that the presence of counterfeits lowers the firm's profit, but may induce the firm to raise its price. Third, we show that the presence of counterfeits has mixed effects on social welfare; only when the wealth inequality among consumers is moderate, will the social welfare increase. Fourth, we demonstrate that increasing the counterfeiter market entry cost does not necessarily improve social welfare despite that it can help eradicate counterfeiters on the market. Managerial implications: Our analysis shows that status signaling and wealth inequality are the key drivers of the strategic interactions between the firm and the counterfeiter as well as the social welfare outcome. The findings from our analysis offer some initial guidelines to managers of luxury brands and e-commerce platforms in addressing the non-deceptive counterfeit problem.

This article develops a method to automatically discover and quantify human-interpretable visual characteristics directly from product image data. The method is generative, and can create new visual designs spanning the space of visual characteristics. It builds on disentanglement methods in deep learning using variational autoencoders, which aim to discover underlying statistically independent and interpretable visual characteristics of an object. The impossibility theorem in the deep learning literature indicates that supervision with ground truth characteristics would be required to obtain unique disentangled representations. However, these are typically unknown in real world applications, and are in fact exactly the characteristics we want to discover. Extant machine learning methods require ground truth labels for each visual characteristic, resulting in a task requiring human evaluation and judgment to both design and operationalize. In contrast, this method postulates the use of readily available product characteristics (such as brand and price) as proxy supervisory signals to enable disentanglement. This method discovers and quantifies human-interpretable and statistically independent characteristics without any specific domain knowledge on the product category. It is applied to a dataset of watches to automatically discover interpretable visual product characteristics, obtain consumer preferences over visual designs, and generate new ideal point designs targeted to specific consumer segments.