Scholarlink

<p>We present a novel statistical model for predicting Scope 1 and 2 emissions for small and medium-sized enterprises (SMEs). Trained on financial transaction data from over 100,000 UK SMEs, the model targets a business segment excluded from formal emissions reporting and often under-engaged in sustainability efforts. By leveraging scalable, objective data, our approach offers an accessible alternative to existing methods that rely on either coarse sectoral averages or detailed, resource-intensive firm-level activity data. In developing the model, we evaluate a range of predictors and find that incorporating industry-level variables beyond basic emission intensity significantly enhances predictive accuracy. We also observe diminishing returns from additional model complexity, reinforcing the value of a parsimonious, low-input design. The final model achieves RSQ values of 0.89 for Scope 1 and 0.72 for Scope 2, improves accuracy by up to 50% compared to sector-level estimates, and performs reliably on out-of-sample data. Our findings provide a simpler approach to emissions estimation for SMEs, supporting broader climate engagement among smaller actors.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Bridging the SME reporting gap: A new model for predicting Scope 1 and 2 emissions

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年9月23日

查看全文
  • Alec Phillpotts,
  • Anne Owen,
  • Jonathan Norman,
  • Anna Trendl,
  • John Gathergood,
  • Norbert Jobst,
  • David Leake

摘要

We present a novel statistical model for predicting Scope 1 and 2 emissions for small and medium-sized enterprises (SMEs). Trained on financial transaction data from over 100,000 UK SMEs, the model targets a business segment excluded from formal emissions reporting and often under-engaged in sustainability efforts. By leveraging scalable, objective data, our approach offers an accessible alternative to existing methods that rely on either coarse sectoral averages or detailed, resource-intensive firm-level activity data. In developing the model, we evaluate a range of predictors and find that incorporating industry-level variables beyond basic emission intensity significantly enhances predictive accuracy. We also observe diminishing returns from additional model complexity, reinforcing the value of a parsimonious, low-input design. The final model achieves RSQ values of 0.89 for Scope 1 and 0.72 for Scope 2, improves accuracy by up to 50% compared to sector-level estimates, and performs reliably on out-of-sample data. Our findings provide a simpler approach to emissions estimation for SMEs, supporting broader climate engagement among smaller actors.