SFDR Data Requirements

When You Need This

The Sustainable Finance Disclosure Regulation (EU 2019/2088) requires financial market participants to disclose sustainability risks and impacts. As a data engineer building ESG data pipelines, you need to know which data points to collect, compute, and publish.

Article Classification (Pipeline Perspective)

ArticleFund TypeData RequirementPipeline Impact
Article 6No sustainability claimsBasic risk disclosureMinimal — text-only disclosures
Article 8Promotes E/S characteristicsPAI indicators for promoted characteristicsModerate — compute selected PAIs
Article 9Sustainable investment objectiveFull PAI disclosure + positive contribution proofHeavy — compute all mandatory PAIs + taxonomy alignment

Principal Adverse Impact (PAI) Indicators

Mandatory Indicators (Table 1 of SFDR RTS)

Quote

“What gets measured gets managed — and what gets disclosed gets scrutinized.”

Hiro Mizuno (former CIO, Japan Government Pension Investment Fund)

#IndicatorData NeededVendor SourceComputation
1GHG emissions (Scope 1, 2, 3)tonnes CO2e per companyMSCI, Sustainalytics, CDPWACI = SUM(weight * emissions / revenue)
2Carbon footprinttonnes CO2e per EUR investedSame + EVICCF = SUM(weight * emissions / evic) * portfolio_value
3GHG intensity of investee companiestonnes CO2e / EUR revenueSameWeighted average
4Exposure to fossil fuels% revenue from fossil fuelsMSCI, BloombergBinary flag + revenue share
5Non-renewable energy share% energy consumptionCDP, MSCIWeighted average
6Energy consumption intensityGWh / EUR revenue per sectorCDPSector-level calculation
7Biodiversity impactOperations near sensitive areasMSCI, SustainalyticsBinary flag
8Water emissionstonnes dischargedCDPWeighted average
9Hazardous wastetonnes generatedCDP, MSCIWeighted average
10UNGC/OECD violationsNumber of violationsMSCI, ISSBinary flag per company
11Gender pay gap% differenceCompany reportsWeighted average
12Board gender diversity% female directorsBloomberg, MSCIWeighted average
13Controversial weapons exposureBinaryMSCI, SustainalyticsBinary flag
14GHG intensity of sovereignstonnes CO2e / GDPWorld Bank, UNFCCCDirect lookup

Data Pipeline for PAI Computation

The PAI computation sits on top of the normalized ESG scores:

ESG vendor data (raw) → Normalize (0-100 scale) → PAI calculation → SFDR disclosure report

Key Computation: WACI

def compute_waci(weights: pd.DataFrame, emissions: pd.DataFrame) -> float:
    """Weighted Average Carbon Intensity (PAI Indicator 3).
    WACI = SUM(weight_i * scope1_2_emissions_i / revenue_i)
    """
    merged = weights.merge(emissions, on='instrument_isin')
    merged['carbon_intensity'] = merged['scope1_2_tonnes'] / merged['revenue_eur_millions']
    return (merged['weight_pct'] * merged['carbon_intensity']).sum()

Data Vendor Mapping to PAI Indicators

PAI IndicatorPrimary VendorBackup VendorCoverage
GHG Scope 1+2MSCI ESGCDP~85% of large caps
GHG Scope 3CDPMSCI (estimated)~40% reported, rest estimated
Fossil fuel revenueMSCIBloomberg~90%
UNGC violationsMSCIISS~95%
Board diversityBloombergMSCI~90%
Controversial weaponsMSCISustainalytics~98%

Coverage Gaps

Scope 3 emissions and some social indicators have low reported coverage. Vendors fill gaps with estimates. Your pipeline must track whether a value is reported or estimated — SFDR requires disclosure of estimation methodology. LLM extraction pipelines can help parse unstructured sustainability reports to fill these gaps with source-attributed data.

ISS & STOXX Glossary