Integrum ESG Core Methodologies

Data

Data gathering and validation

Our ratings are built from what companies publish. Three layers turn those disclosures into data that can be scored and checked.

1. Proprietary AI models More than 200 machine learning models, each built and trained for more than five years on a specific ESG metric, read company reports and extract the relevant data.
2. Systematic data cleaning A data-centric AI platform using semi-supervised learning reviews millions of datapoints every month, detects inconsistencies and refines the data the models are trained on.
3. Expert review An ESG analyst reviews every datapoint before it is surfaced in the Glass Box. Each correction feeds back into the models, so their accuracy improves over time.

This combination lets us update our data within 10 days of a company publishing a new report. For more on how AI is changing ESG data collection, read AI as ESG Infrastructure. Clients can apply the same models to their own companies, sources and data points through Golden Retriever.

IFRS S1

Sustainability: sector-specific metrics

Integrum ESG is a licensee of the ISSB Standards, based on the SASB Materiality Map. The Map sets out the most important sustainability issues for each subsector and was developed over five years with input from multiple stakeholder groups. Our Sustainability dimension covers 26 metric categories and 44 sub-metrics, and each company is scored only on those material to its subsector.

• Qualitative metrics assess a company’s awareness of an issue and receive an Awareness Score.
• Quantitative metrics assess both awareness and performance, and receive an Awareness Score and a Performance Score, which are averaged.

Both scores run from 0 to 4. The definition of every score level is set out in the Integrum ESG Scoring Framework.

IFRS S2

Climate Risk: climate-related metrics

IFRS S2 is built on the recommendations of the Taskforce on Climate-related Financial Disclosures (TCFD), now maintained as Standards by the ISSB, which Integrum ESG formally licenses. Four metrics together form the Climate Risk grade, which is reported separately from the ESG grade.

Each Performance Score ranks the company in quartiles against its peers. Our TCFD Reporting and Climate VaR Methodology explains how we calculate Climate VaR and the core climate metrics for funds, and our guide to TCFD reporting in 2027 covers the current FCA rules.

Governance

Governance metrics

The SASB Standards focus on sustainability rather than governance, so we assess governance using the framework of our partner Minerva Analytics. It covers 9 metrics and 39 sub-metrics, applied to every company, and Minerva captures more than 1,000 governance data items per company from public disclosures.

Governance carries many more sub-metrics than sustainability because, like many investment firms, we regard governance as highly price-sensitive, and good environmental and social performance is hard to sustain without it. Many governance metrics are qualitative, and some use fixed number boundaries in place of peer quartiles, for example the vesting period of long-term incentive plans. As with every other metric, each governance score comes with a Glass Box showing why it was awarded and the company data behind it.

Impact

Impact metrics and the UN SDGs

We assess how each company is advancing the UN Sustainable Development Goals using the Cambridge Impact Framework, developed at the University of Cambridge Institute for Sustainability Leadership with a group of impact-focused fund managers. It uses six themes, each represented by one metric that most companies disclose.

To score highly, a company must disclose the relevant data, set quantifiable targets for each theme and show awareness of the specific SDGs. Impact Performance Scores are benchmarked against our full coverage universe. The Impact score is separate from the ESG score and does not feed into it. Our SDG Impact from Operations Methodology sets out each theme in full.

Grades

Company ESG grade

1. Weight each metric Each material metric carries equal weight by default, because the ISSB does not treat any issue it deems material for an industry as more important than another. Clients can change the weightings for their own use.
2. Build the dimension scores Each metric’s score is multiplied by its weight, and the weighted scores are summed into Sustainability, Governance and Impact scores on the 0 to 4 scale.
3. Combine into the overall score By default, the overall ESG score is the average of the Sustainability and Governance scores.
4. Convert to a grade Grades from A to E are set by standard deviation bands around 2, the midpoint of the scale. A C covers scores within 0.5 standard deviations of 2; B and D cover scores between 0.5 and 1.5 standard deviations above or below; A and E cover scores beyond 1.5 standard deviations.

On the Platform, tooltips and a Scoring Logic page show the Awareness and Performance rules behind any metric or sub-metric. Grade bands, fund and portfolio scores and worked examples are set out in the Integrum ESG Scoring Framework, and the company view is described in Company ESG Scores and Glass Box Methodology.

Temperature

Implied Temperature Rise

The Implied Temperature Rise (ITR) is the level of global warming, in degrees Celsius, that would result if every company behaved in the same way. We calculate it for every fund and every company in an uploaded portfolio, by comparing the company’s emissions trajectory with a global carbon budget:

ITR = α × (Carbon Intensity / (Revenue Intensity × 1bn)) × Global GDP × θ × Years to 2100 + β

• Carbon Intensity: the company’s latest annual emissions per $1m of enterprise value, or market capitalisation.
• Revenue Intensity: annual revenue divided by enterprise value, or market capitalisation.
• α, β and θ: constants from a University of Cambridge model.
• Global GDP: a fixed global GDP assumption used for scaling.
• Years to 2100: the number of years from the current year to 2100.

For funds, holdings are looked through and reweighted before the portfolio calculation. If BP makes up 20% of a fund and that fund makes up 10% of a fund of funds, BP carries a weight of 2%. Our Temperature Score and Implied Temperature Rise Methodology covers the four stages, the Paris alignment labels and custom emissions reduction rates.

SFDR

SFDR and principal adverse impacts

We map every holding in a portfolio against the 14 mandatory principal adverse impact (PAI) indicators under SFDR and assigns the portfolio an Article classification. Many PAIs correspond directly to data we already hold. Where the exact data SFDR specifies is not available, we use relevant proxy data, as Article 7(2) of the SFDR Regulatory Technical Standards allows. For example, few companies disclose an unadjusted gender pay gap, so we use workforce gender diversity and alignment with SDG 5 as proxies for whether a company treats women fairly in the workplace.

PAI colour system

Article classification at company level

We assign an Article 6, 8 or 9 alignment to each holding as a bridge between the PAIs and the fund classification.

• Article 8: no more than four PAIs marked yellow or red combined, with clear evidence of good governance practices.
• Article 9: no PAIs marked yellow or red, support for the relevant sustainability objectives and clear evidence of good governance practices.
• Article 6: companies that meet neither set of criteria.

Every company-level PAI has a Glass Box, which shows the underlying data and the reason for the assessment. Our SFDR Classification and PAI Methodology sets out the full rules, including how funds are classified.

Sentiment

Real-time ESG sentiment

Disclosure-based data looks backwards, so we supplement it with real-time sentiment analysis. Our AI model monitors more than 47,000 news and social media sources, filters each story for relevance to the company and to our ESG metrics and classifies it as positive, negative or neutral.

The net sentiment score covers the past seven days. It ignores neutral mentions and is calculated as the percentage of the remaining mentions that are positive minus the percentage that are negative. Data updates every 15 minutes and is available as an API feed, so clients can set their own time period. Sentiment is reported separately from our ESG grades. Our ESG Controversies and Sentiment Methodology sets out the detail.

Consensus

Consensus ESG scores

Our consensus score, provided by CSRHub, shows how the wider market rates a company. It combines assessments from more than 900 sources into a 0 to 100 rating, a percentile ranking against industry and country peers and an A to E grade, and is reported separately from our in-house ratings. The Consensus ESG Score Methodology sets out how it is built.

Related methodologies

• Integrum ESG Scoring Framework
• Company ESG Scores and Glass Box Methodology
• Consensus ESG Score Methodology
• Controversial Activities Methodology
• ESG Controversies and Sentiment Methodology
• Private Markets ESG Data Capture Methodology
• EDCI Data Collection and Reporting Methodology
• SFDR Classification and PAI Methodology
• EU Sustainable Investment Audit Methodology
• EU Taxonomy Alignment Methodology
• TCFD Reporting and Climate VaR Methodology
• Portfolio Footprint Calculator Methodology
• Fund ESG Ratings Methodology
• Golden Retriever Data Extraction Methodology
• Heatstrips and Peer Benchmarking Methodology
• Proxy ESG Scores Methodology
• ESG Screening Methodology
• SDG Impact from Operations Methodology
• Sovereign ESG Scores Methodology
• Temperature Score and Implied Temperature Rise Methodology
• SDG Revenue Alignment Methodology