While the value of using social media information has been established in multiple business contexts, the field of operations and supply chain management have not yet explored the possibilities it offers in improving firms' operational decisions. This study attempts to do that by empirically studying whether using publicly available social media information can improve the accuracy of daily sales forecasts.We collaborated with an online apparel retailer to assemble a dataset that combines (1) detailed internal operational information, including data on sales, advertising, and promotions, as well as (2) publicly available social media information obtained from Facebook. We implement a variety of machine learning methods to forecast daily sales. We find that using social media information results in statistically significant improvements in the out‐of‐sample accuracy of the forecasts, with relative improvements ranging from 12.85% to 23.23% over different forecast horizons. We also demonstrate that nonlinear boosting models with feature selection, such as random forests, perform significantly better than traditional linear models. The best‐performing method (random forest) yields an out‐of‐sample MAPE of 7.21% when not using social media information and 5.73% when using social media information is used. In both cases, this significantly improves the accuracy of the company's internal forecasts (a MAPE of 11.97%). Combining these empirical results, we provide recommendations for forecasting sales in general as well as with social media information.
Recent research has found widespread discrimination by hosts against guests of certain races in online marketplaces. In this paper, we explore ways to reduce such discrimination using online reputation systems. We conducted four randomized field experiments among 1,801 hosts on Airbnb by creating fictitious guest accounts and sending accommodation requests to them. We find that requests from guests with African American–sounding names are 19.2 percentage points less likely to be accepted than those with white-sounding names. However, a positive review posted on a guest’s page significantly reduces discrimination: when guest accounts receive a positive review, the acceptance rates of guest accounts with white- and African American–sounding names are statistically indistinguishable. We further show that a nonpositive review and a blank review without any content can also help attenuate discrimination, but self-claimed information on tidiness and friendliness cannot reduce discrimination, which indicates the importance of encouraging credible peer-generated reviews. Our results offer direct and clear guidance for sharing-economy platforms to reduce discrimination. This paper was accepted by Vishal Gaur, operations management.
The Hospital Readmissions Reduction Program (HRRP), a part of the US Patient Protection and Affordable Care Act, requires the Centers for Medicare and Medicaid Services to penalize hospitals with excess readmissions. We take an economic and operational (patient flow) perspective to analyze the effectiveness of this policy in encouraging hospitals to reduce readmissions. We introduce a single-hospital model to capture the dependence of a hospital's readmission-reduction decision on various hospital characteristics. We derive comparative statics that predict how changes in hospital characteristics impact the hospital's readmissionreduction decision. We then proceed to develop a game-theoretic model that captures the competition between hospitals introduced by the HRRP policy's benchmarking mechanism. We provide bounds that apply to any equilibrium of the game and show that the comparative statics derived from the single-hospital model remain valid after the introduction of competition. Importantly, the comparison of the single-hospitals and multi-hospital models shows that, while the competition among hospitals often encourages more hospitals to reduce readmissions, it can only increase the number of "worst offenders," which are hospitals that prefer paying penalties over reducing readmissions in any equilibrium. We calibrate our model with a dataset of hospitals in California which allows us to quantify the results and insights derived from the model. Last, we validate our model with recent hospitals' performance data collected since the policy was implemented.
In reward-based crowdfunding, creators of entrepreneurial projects solicit capital from potential consumers to reach a funding goal and offer future products/services in return. The authors examine consumers’ contribution patterns using a novel data set of 28,591 projects collected at 30-minute resolution from Kickstarter. Extending prior research that assumes that economic considerations (e.g., project quality, campaign success likelihood) drive backers’ decisions, the authors provide the cleanest field evidence so far that consumers also have prosocial motives to help creators reach their funding goals. They find that projects collect funding faster right before (vs. right after) meeting their funding goals because consumers not only are more likely to fund projects but also contribute greater amounts of money prior to goal attainment. This effect is amplified when the nature of a project tends to evoke consumers’ prosocial motivation and when a project’s creator is a single person. These results suggest that consumers’ prosocial motives not only play a role in reward-based crowdfunding but also can outweigh the opposing effects of economic factors including rational herding and certainty about campaign success.
Many online retailers provide real-time inventory availability information. Customers can learn from the inventory level and update their beliefs about the product. Thus, consumer purchasing behavior may be impacted by the availability information. Based on a unique setting from Amazon lightning deals, which displays the percentage of inventory consumed in real time, we explore whether and how consumers learn from inventory availability information. Identifying the effect of learning on consumer decisions has been a notoriously difficult empirical question because of endogeneity concerns. We address this issue by running two randomized field experiments on Amazon in which we create exogenous shocks on the inventory availability information for a random subset of Amazon lightning deals. In addition, we track the dynamic purchasing behavior and inventory information for 23,665 lightning deals offered by Amazon and use their panel structure to further explore the relative effect of learning. We find evidence of consumers learning from inventory information: a decrease in product availability causally attracts more sales in the future; in particular, a 10% increase in past claims leads to a 2.08% increase in cart add-ins in the next hour. Moreover, we show that buyers use observable product characteristics to moderate their inferences when learning from others; a deep discount weakens the learning momentum, whereas a good product rating amplifies the learning momentum. This paper was accepted by Serguei Netessine, operations management.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.