Ask an underwriter how confident they are in a corporate risk score, and most will
hedge. Not because the model is wrong, but because they know how much of it is
inferred rather than confirmed.
That distinction matters more this year than ever. EIOPA’s own climate risk
commentary this April put it plainly: better data leads to better pricing, and better
pricing leads to better risk management. The gap starts with where the underlying data
actually comes from.
Two Very Different Starting Points
A risk model can be built up in layers. The first layer is modelled: sector averages and
nearest-neighbour comparisons applied where a company hasn’t yet been directly
evidenced. It’s a reasonable estimate for coverage at scale, but it’s still an estimate for
that company.
The next layer replaces the estimate with the company’s own reported evidence:
disclosures, sustainability reports, and quantitative data such as GHG emissions
figures.
The difference isn’t “good data” versus “no data.” It’s an estimate applied before a
company has been evidenced, versus that company’s own reported evidence once it
has.
A Worked Illustration
Take a mid-market manufacturing company being assessed for underwriting risk.
Before it has been directly evidenced, the picture is built from modelled estimates:
• Sector-average emissions intensity applied in place of company-specific figures
• Governance score extrapolated via similarity to comparable firms in the same
industry
• No confirmation of physical site exposure or supply chain concentration
Once that company’s own reported disclosures are brought in, the risk view can shift
substantially:
• Reported emissions data replaces the sector average, often revealing the
company sits above or below its peer group
• Governance indicators come from the company’s own disclosures rather than
comparable firms
• Site-level and supply chain exposure becomes visible, rather than assumed
The risk score doesn’t just become more accurate. It becomes traceable to a source,
which matters as much to a supervisor as to a pricing actuary.
Why This Distinction Belongs in ORSA and Underwriting Conversations Now
Solvency II, ORSA, and SFDR disclosures increasingly ask insurers to show their
working, not just their conclusions. A score resting entirely on modelled estimates is
harder to defend when a supervisor or reinsurer asks where a number came from.
This is where a tiered data model earns its place. Moving a counterparty from modelled
estimates to its own reported disclosures doesn’t require re-underwriting the whole
book at once. It means knowing which parts of a risk view are the company’s own
evidence, and which are still modelled.
What This Means for Insurers and Corporates
Modelled estimates still have a place, especially for portfolio-wide coverage at first
assessment. The question is whether you know, counterparty by counterparty, which
figures are the company’s own reported evidence and which are still modelled.
Talk to Our Team
Public disclosure analysis was built precisely to answer that question, mapping 120
SASB-aligned metrics across sectors and combining reported disclosures with
quantitative and verified data, so insurers and corporates can see exactly which parts
of their risk view rest on a company’s own evidence and which are still modelled.
Talk to our team to see your own portfolio through that lens.
