补贴
现金流
排名(信息检索)
标杆管理
潜在Dirichlet分配
计量经济学
变化(天文学)
环境经济学
经济
业务
产业组织
财务
计算机科学
营销
物理
机器学习
天体物理学
主题模型
自然语言处理
市场经济
标识
DOI:10.1287/msom.2022.1115
摘要
Problem definition: Carbon abatement opportunities are diverse, making it difficult to classify them. Do latent classes of carbon abatement opportunities exist and is there a type that is financially and environmentally superior? Methodology/results: In this study, we classify 16,525 implemented carbon abatement projects using text analysis. We benchmark our clustering method to the latent Dirichlet allocation model and verify our classifications using a crowd-sourcing platform. We then compare the payback period, financial hurdle (measured in upfront cost), savings, and carbon emissions reduction by type. Our results show that latent classes exist, and they statistically differ in the metrics we examine. Our regression results show that the type of project explains more of the variation in the financial and environmental outcomes than the firm-level financial controls we included. We find that liquidity (measured using cash-to-asset and current ratios) is associated with the number of reported projects, but the magnitude and direction varies by type. Our extension shows that marginal abatement costs statistically differ by type with a few exceptions. Lastly, we show that our classification is robust to sector-level variation. Managerial implications: Although the results show that no single type of opportunity dominates in all four metrics, our classification provides a ranking of the types firms should pursue depending on their goals. Our results suggest that firms likely place different weights across these four metrics. This means that policies targeted at making investment costs more attractive (e.g., subsidies or better financing) may not have the same impact on firms that put more weight on savings compared with those more sensitive to costs. A classification of opportunities can contribute toward understanding whether a unifying theory or pattern across carbon abatement activities may exist or not.
科研通智能强力驱动
Strongly Powered by AbleSci AI