Research Corpus
~400 peer-reviewed academic papers, synthesized into findings you can act on. Every claim grounded in published research.
Knowledge Dashboard
The Research tab presents a Knowledge Dashboard — synthesized findings from the corpus, not a list of papers. The dashboard leads with answers:
- Key Findings — the strongest evidence across the corpus, presented as narrative cards with effect size ranges and confidence levels
- By Hazard — findings grouped by climate driver (flood, heat, drought, sea level rise, carbon transition, disclosure/regulation)
- By Asset Class — findings grouped by market segment (public equity, corporate credit, insurance/ILS, muni/sovereign)
- Evidence Landscape — a matrix showing where evidence is strong, moderate, emerging, or absent across hazard–outcome pairs
Each finding shows an effect size range, confidence level (strong, moderate, or emerging), and the hazard, asset class, and transmission channel it applies to. No paper titles, no authors, no DOIs — just what the research says and how confident we are.
What's Included
The CPI research corpus contains peer-reviewed academic papers focused on climate risk and financial valuation:
- ~400 papers from top journals (Journal of Finance, Real Estate Economics, etc.)
- Full text indexed and searchable by the chat agent
- Extracted figures and tables available for citation
- Structured metadata — authors, journal, publication date, DOI
- Semantic search — the agent finds papers by concept, not just keywords
Topics Covered
| Category | Examples |
|---|---|
| Physical Risk | Flood, heat, wildfire, hurricane, sea level rise |
| Transition Risk | Carbon pricing, stranded assets, policy changes |
| Asset Classes | Residential, commercial, REITs, municipal bonds |
| Financial Metrics | Price effects, cap rates, NOI, insurance costs |
| Geography | US coastal, European, global markets |
How Search Works
When you ask a question in chat, the CPI agent performs semantic search over the corpus:
- Query understanding — Your question is analyzed for key concepts
- Semantic matching — Papers are ranked by conceptual similarity, not just keyword matches
- Synthesis — Relevant findings are synthesized into a coherent answer
- Citation — Sources are cited so you can click through and verify
This means you can ask questions in natural language:
- “What's the price effect of flood risk on residential properties?”
- “How does extreme heat affect commercial real estate values?”
- “What do the academic papers say about hurricane risk and insurance costs?”