Our Ratings Methodology
EthicsGrade Core Functions
At EthicsGrade we have the only comprehensive data set in the world on the ESG issues relating to corporate digital responsibility (CDR). The methodology behind these ratings stems from the team at EthicsGrade which is comprised of two core functions. The first is a traditional research analyst team, and the second is an automation engineering function.
Traditional research analysis
While other methods to automate ESG research suffers from a poor signal-to-noise ratio which results in the consequent data being insufficient for investment quantitative analysis, our belief is that by having a traditional research team at our core we are able to create time leverage for our analysts as opposed to simply amplifying data, including data quality weaknesses.
Automation engineering
The automation engineering team continuously monitors the research analyst process as well as the completion of InsideView surveys and as a result looks for opportunities to fully automate answers to some of the questions in the model. Where this has been achieved, the opportunity to correct and amend the answers is still offered to the company in coverage, as this also creates opportunities to further train and refine the process.
EthicsGrade
Research Model
Our approach to ESG ratings is simple yet powerful. We are building a 'data-driven materiality platform' that understands the specific risks that each organisation faces – and compares the maturity of their governance against their peers and wider best-practice.
1. GATHER DATA & EVIDENCE
Our analyst team operates from a research 'model', which is a dynamic set of questions related to each company under coverage. The relevance or materiality of each question to each company varies, and thus we use data from stakeholders collected via our website to ascertain which questions are of the greatest relevance to any company.
2. SELECT RELEVANT QUESTIONS
The research team records the evidence found, and the extent to which each data point confirms an answer to a particular question. At the end of each quarter, the score for each company is calculated. The calculation is broken down by 'dimension' a shorthand for categorisation of the various themes in the model, although by no means these dimensional headings should indicate any preference for grouping or weighting of scores. Instead, each question is continuously weighted towards what a reflection of the stakeholder preferences is collected via the EthicsGrade website.
3. CALCULATE THE SCORE
Preference is given to authoritative sources such as the company's published sustainability or ESG report or other formal communications from the company. In some cases, press releases from vendors or partners are taken into consideration, especially when they confirm the existence of a particular set of controls or governance. At present the EthicsGrade model comprises 340 questions and due to increases in scope and depth of coverage, it is growing at approximately 25% per year.
4. QUALITY REVIEW & PUBLISH
Once all company scores have been calculated and grades allocated (according to the published scale), the analyst team performs a 'red-team' exercise where other members of the group are invited to challenge particular answers and company results. This process may result in a change to the answer determined for a question relating to the company, a change in the weighting applied to the question, or might cause the introduction of new questions which further refine the level of detail and nuance which is encoded in the EthicsGrade data-set.
EthicsGrade Tools
InsideView survey
Once the scores for all companies in research coverage have been completed for that quarter, the scorecards are produced and sent to each company for review before publication on the EthicsGrade website and in the data feed which licensed users (academics, journalists, professional investors, and customers of companies) receive. Each company is invited to participate in the InsideView survey which dynamically assigns the most material questions to that company for further answers or to provide non-public data which refines the ESG rating of that company.
Prediction engine
For questions which are not deemed to be sufficiently material for each company, and for companies which are graded but fall outside of EthicsGrade's research coverage, the assessment is automated using EthicsGrade's prediction engine. The prediction engine uses data that it has been trained on which includes analyst research to predict the likelihood of any potential answer to every question in the EthicsGrade model. Predictions are similarly red-teamed and randomly selected for manual review, thereby creating the process of back-testing which is important to further train and develop the prediction engine.
