Hiding in the Herd: The Product Recall Clustering Phenomenon

• 成果类型: Article
• 署名作者: Mukherjee, Ujjal K.; Ball, George P.; Wowak, Kaitlin D.; Natarajan, Karthik, V; Miller, Jason W.
• 署名单位: University of Illinois System; University of Illinois Urbana-Champaign; Indiana University System; IU Kelley School of Business; Indiana University Bloomington; University of Notre Dame; University of Minnesota System; University of Minnesota Twin Cities; Michigan State University; Michigan State University's Broad College of Business
• 刊物名称: M&SOM-MANUFACTURING & SERVICE OPERATIONS MANAGEMENT
• ISSN/ISSBN: 1523-4614
• DOI: 10.1287/msom.2020.0937
• 发表日期: 2022
• 页码: 392-410
• 关键词: recall clustering; herding theory; nash stopping-time games; hawkes process model; event study

摘要

Problem definition: Product recalls have serious consequences for firms and consumers. Little is known, however, about how firms manage the timing of recalls and how this timing influences the financial consequences of recalls. In this study, we provide evidence for a previously unknown phenomenon: recall clustering, a collection of recalls within close temporal proximity in which a leading recall (the first recall in a cluster) excites following recalls (subsequent recalls in a cluster). We also investigate how the stock market penalizes firms differently depending upon their position within the recall cluster. Academic/practical relevance: By demonstrating that auto firms cluster their recalls and that the market penalizes firms differently based on the position of a recall within a cluster, we contribute to the literature that investigates recall timing and stock market event studies and provide guidance for regulators who oversee auto recalls and managers who make recall decisions. Methodology: We first develop analytical predictions using a dynamic game theoretic model to motivate our hypotheses. We then examine empirical support for our hypotheses by analyzing 3,117 auto recalls across 48 years using a Hawkes process model. Hawkes process models are designed to examine self-excitation of events across time and can be used to investigate recall clustering, while categorizing recalls as leading or following within a cluster. Finally, we use the leading and following recall designations obtained from the Hawkes process model in an event study to examine how the stock market effects of a recall vary depending on its position within a cluster. Results: We find that 73% of recalls occur in clusters, and they form after a 16-day gap in recall announcements. On average, clusters last for 34 days and are comprised of 7.6 following recalls announced after the leading recall. Leading recalls are associated with as high as a 67% larger stock market penalty than following recalls. Further, we find that the stock market benefit realized by a following recall weakens as the time since the leading recall increases and that the stock market penalty faced by a leading recall grows as the time since the end of the last cluster increases. Managerial Implications: Our findings lead to a key implication for regulators who oversee auto recalls by demonstrating evidence of recall clustering and the underlying stock market effects that are attributable to it. We provide a cost-neutral policy recommendation for the National Highway Traffic and Safety Administration that should limit recall clustering.