Sustainability Raw Data Disclosed & Modelled

ESG metrics for up to 44 impact KPIs for 50 leading companies, with flags for whether the data is from company disclosures or GIST Impact models. The full data set is available for over 13,000 companies worldwide.

Description

Raw data is crawled from publicly available company disclosures using our cognitive search engine. The data undergoes validation by our team of expert analysts to identify, verify and document outliers. Following reprocessing and data appending, the data undergoes algorithmic assurance before final approval by team leads specialising in each area of impact. The combination of human and machine quality control delivers a high level of confidence in the accuracy of the data.

Disclosure is incomplete even for GHG emissions, which is the most standardised non-financial disclosure. The lack of data is significantly higher in emerging markets compared with developed and certain sectors particularly lacking disclosure. Water usage and waste generation data are much sparser than GHG emissions.

Benchmark-based models are commonly used in the industry. The benchmarking approach uses the average emissions of the companies belonging to a particular business activity to predict the emissions of companies with no disclosures, by extrapolating the emissions based on revenues. Estimations are linearly proportional to the revenue generated by the company. This approach ignores that most large business have revenues from multiple sectors and that location is particularly important when assessing impact. Statistically, each estimated value has the same standard deviation, which leads to very limited co-relation between estimated and actual values.

GIST Impact’s machine learning approach to modelled data takes into consideration unique financial and non-financial data points for each company. A unique list of companies is identified to create the nearest neighbour group and then an average value is generated for extrapolation. Six different unique operational parameters are taken into consideration to identify the peer group.

The model accounts for multiple business activities and the peer group accurately captures the variability from operations, location, and performance, thereby reducing the deviation between actual emissions and estimates.

Customers have the option of access to analysts for questions and clarifications of the data.

Data

Meta-data Fields:
• ISIN (licence required)
• ISO Code
• Company Name
• Reporting year
Impact Data:
1. Scope 1 emissions
2. Scope 2 emissions – location based
3. Scope 2 emissions – market based
4. Total Scope 2 emissions
5. Total GHG emissions
6. Total energy consumption
7. Coal
8. Diesel
9. Motor Gasoline
10. Natural Gas
11. Particulate matter
12. Oxides of Nitrogen
13. Oxides of Sulphur
14. Air Cadmium
15. Air Mercury
16. Total water withdrawal
17. Total freshwater withdrawal
18. Total water consumption
19. Total water discharge, wastewater generation
20. Recycled water, reuse water, treated water
21. Chemical oxygen demand
22. Total non-hazardous waste generated
23. Total hazardous waste generated
24. Total waste generated
25. Waste incinerated
26. Waste composted
27. Waste to landfill
28. Disposed – non-hazardous waste
29. Disposed – hazardous waste
30. Disposed – total waste
31. Recovered – non-hazardous waste
32. Recovered – hazardous waste
33. Total waste recovered / recycled
34. Fly ash
35. Construction debris
36. Overburden in mining
37. Number of employees
38. Number of female employees
39. Number of male employees
40. Percentage of female employees
41. Average age
42. Annual revenue
43. Start date of accounting period
44. End date of accounting period

Flags are provided to show whether impact data is from disclosed sources or modelled.

搜集汇总

背景与挑战

背景概述

该数据集提供全球超过13,000家公司的ESG指标，涵盖44个关键影响指标，并标注数据来源为公司披露或GIST Impact模型。它通过结合人工验证与机器学习方法进行质量控制，以应对非财务披露不完整的问题，特别是在新兴市场和特定行业。