How Util Collects Data
Util depends on high quality data to support its methodology. Any data that we use must align with our Principles of Good Data:
- Data must be objective - audited accounts are a good source of data. A company detailing its commitment to the community is not;
Data must be universal – all companies have an environmental footprint, creating waste and using water, even if this footprint is very light. We do not measure company or theme specific metrics and key performance indicators.
Data cannot be gamed – by the company or other interested actors. We seek external data on employee satisfaction, above company created employee satisfaction surveys.
Data must be an output or outcome of an activity – policies and commitments are not good proxies of actual value creation, dollars spent on training and development and employee sentiment analysis are.
Data must be automated – we must be able to gather data for our universe of listed companies with no manual work. Building natural language processing models that match companies and their products to UN Sustainable Development Goals is scalable and automated, manually searching through LinkedIn to find a company’s countries of employment is not.
In order to gather data of this nature, feeding Util’s bottom line analysis, we look to three distinct data sources. We start with company disclosure – audited and objective data found within a company’s financial statements, annual reports and sustainability overviews. We are specifically looking for environmental disclosure (Carbon, Waste and Water), financial fundamentals (revenue, salaries, spend on research and development) and employee items (amount spent on training and development). When company disclosure is not sufficient, we seek to build data relationships with external providers who deliver aggregated data on stakeholder specific areas. For example, we use Glassdoor and Good Places to Work for employee satisfaction and Equileap for equality analysis. Finally, when we cannot find data sources that apply to our rules, detailed above, we create our own frameworks for gathering data. Examples include the matching of a company’s products and services to UN Sustainable Development Goals, using Google Trends to better proxy a company’s revenue breakdown by country.
The universe of companies that we can analyse extends to 65,000 listed equities. However, disclosure laws vary depending on listing and jurisdiction and therefore we have a methodology to proxy data that isn’t disclosed by a company. We split non-disclosure into three categories: fundamental, negative and positive, with varying approaches:
Fundamental: data contained in financial statements that underpins our methodology, including number of employees and amount spent on salaries. In these cases, we take the median of industry peers, weighted against an appropriate constant (for example, total expenditure or total revenue) to arrive at an appropriate proxy.
Negative: data relating to value destruction, usually found within the Environment stakeholder methodology, for example Scope 1 & 2 carbon emissions. In these cases, we take the worst performer within an industry, weighted against expenditure or revenue, applying this ratio to the non-disclosing company in question. We then follow up with the company, seeking disclosure.
Positive: data relating to value creation, for example volunteer hours or amount spent on training and development. In these cases, only disclosure gets rewarded – if a company fails to disclose, no positive value is awarded.