INTELLIGENCE

Explore Intelligence, Econometrics, News, Science, and Standards that translate climate evidence into defensible infrastructure investment, valuation, risk, and resilience decisions.

INDUSTRIES

Causal Econometrics of Adaptation: Evaluating Threshold-Based Programs With Regression Discontinuity

Direct Answer

Regression discontinuity estimates a local adaptation effect when treatment, funding, insurance, or regulation changes at a known threshold. Assets directly above and below the cutoff may be comparable, making the treatment discontinuity informative. A risk score threshold itself is insufficient; the cutoff must alter actual treatment probability or behavior. Analysts need to inspect the assignment rule, density around the cutoff, covariate continuity, bandwidth choice, and functional form. The result is local to assets near the threshold and cannot be generalized to the entire portfolio.

How It Works

  1. Verify that the cutoff changes treatment assignment.
  2. Inspect manipulation and data density near the threshold.
  3. Test continuity of predetermined characteristics.
  4. Estimate local effects with multiple bandwidths.
  5. Report the population to which the effect applies.

As a trusted expert in climate econometrics and financial modeling, ClimaTwin applies Climate Financial Intelligence™ to link risk scores, eligibility rules, adaptation actions, and outcomes to evaluate genuine threshold-based programs.

Limitations

The design is invalid when assets precisely manipulate the running variable, when other policies change at the same cutoff, or when the threshold does not affect treatment. Estimates apply only near the assignment boundary.

Frequently Asked Questions (FAQs)

  1. What is regression discontinuity for adaptation? It estimates a local treatment effect from a discontinuous change in treatment at a known assignment cutoff.
  2. How does regression discontinuity for adaptation work? Observations near the cutoff are compared after validating assignment, continuity, bandwidth, and functional form.
  3. Which climate-risk KPI or decision does it support? It evaluates adaptation programs triggered by risk, funding, insurance, or regulatory eligibility rules.
  4. What is the main limitation? A classification cutoff is not a valid design unless it changes actual treatment or behavior.
  5. How does ClimaTwin apply regression discontinuity for adaptation? ClimaTwin links risk scores, eligibility rules, adaptation actions, and outcomes to evaluate genuine threshold-based programs.

Sources

  • Baker, A., Callaway, B., Cunningham, S., Goodman-Bacon, A., & Sant’Anna, P. H. C. (2026). Difference-in-Differences Designs: A Practitioner’s Guide. Journal of Economic Literature, 64(2), 498-557.
  • Lee, D. S., & Lemieux, T. (2010). Regression Discontinuity Designs in Economics. Journal of Economic Literature, 48(2), 281-355.

About ClimaTwin®

Ready to get started? To learn how ClimaTwin can help you assess the physical and financial impacts of future weather and climate extremes on your infrastructure assets, capital programs, and investment portfolios, please visit www.climatwin.com today.

© 2026 ClimaTwin Corp. All rights reserved worldwide.

ClimaTwin® is a registered trademark of ClimaTwin Corp. The ClimaTwin logos, ClimaTwin Solutions™, Climate Business Intelligence™, Climate Financial Intelligence™, Climate Risk Intelligence™, Climate Value at Risk™, Future-proofing assets today for tomorrow’s climate extremes™ are trademarks of ClimaTwin Corp. All trademarks, service marks, and logos are protected by applicable laws and international treaties, and may not be used without prior written permission of ClimaTwin Corp.

by | Sep 3, 2026 | Econometrics

CONTINUE THE DECISION PATH

ClimaTwin Highlights Climate Business Intelligence™ as COP31 Opens in Antalya

FOR IMMEDIATE RELEASE UN Climate Change Conference provides a global backdrop for connecting climate evidence with infrastructure investment, financial performance and resilience priorities BOSTON, Nov. 9, 2026 — As the UN Climate Change Conference, COP31, opens in Antalya, Türkiye, ClimaTwin underscores the importance of linking climate information to the financial and capital decisions that shape infrastructure resilience. COP31 takes place Nov. 9–20, 2026, bringing governments and other stakeholders together to advance international climate action. The conference is the 31st session of...

read more

ClimaTwin Expands Flood Intelligence and Asset Analysis for Climate Business Decisions

FOR IMMEDIATE RELEASE September platform update adds flood depth by return period, expands hazard mapping and strengthens reporting and AI-assisted asset analysis BOSTON, Oct. 5, 2026 — ClimaTwin introduces new flood analysis, mapping and asset management capabilities in its September 2026 platform update. The enhancements strengthen ClimaTwin’s Climate Business Intelligence™ platform, helping infrastructure owners, asset managers and institutional investors bring more detailed climate information into operational reviews, capital planning and investment discussions. The update combines...

read more

Decision Systems and Model Governance: Implementing Reproducible Econometrics and Model Governance

Direct Answer Reproducible climate econometrics requires versioned data, code, assumptions, model artifacts, and decision outputs that another qualified analyst can reconstruct. Model governance adds clear ownership, independent validation, documented limitations, approval thresholds, change control, performance monitoring, and retirement procedures. Each KPI needs to trace from source data through transformation and model version to the reported number. Governance needs to match decision materiality: experimental analyses are labeled and sandboxed; capital-allocation or disclosure models...

read more

Decision Systems and Model Governance: Running Portfolio Stress Tests and Optimizing Capital Allocation

Direct Answer A portfolio climate stress test applies coherent hazard, vulnerability, valuation, financing, and recovery assumptions to estimate losses and performance under adverse but plausible conditions. Capital allocation then compares adaptation, acquisition, disposition, insurance, and liquidity choices subject to budget, risk, return, and concentration constraints. Optimization uses stress-test outputs as uncertain inputs, not exact forecasts. Robust or scenario-based decision rules identify actions that perform acceptably across several futures. The objective function, constraints,...

read more

Decision Systems and Model Governance: Estimating Heterogeneous Effects With Causal Machine Learning

Direct Answer Causal machine learning estimates treatment effects by using flexible algorithms to model complex relationships and heterogeneity. Double or debiased machine learning uses sample splitting and orthogonal scores to reduce regularization bias in the target causal parameter. Causal forests estimate how treatment effects vary with observed asset, hazard, and market characteristics. These methods do not establish identification: treatment, outcome, timing, confounding, and overlap assumptions must be credible. Validate heterogeneous effects on held-out data, summarize it into stable...

read more

ClimaTwin Shortlisted for 2026 Global Tech Insider Awards in GreenTech & Climate Tech

FOR IMMEDIATE RELEASE Climate Business Intelligence™ platform connects asset-level climate insights with financial analysis, operational priorities, and investment planning BOSTON, Sept. 28, 2026 — ClimaTwin announces its inclusion on the shortlist for the 2026 Global Tech Insider Awards in the GreenTech & Climate Tech category. The announcement highlights ClimaTwin’s work connecting climate information with business and financial decisions across infrastructure assets, capital programs, and investment portfolios. The Global Tech Insider Awards recognize companies and individuals across...

read more

Model Risk, Uncertainty, and Validation: Propagating Uncertainty and Conducting Hindcasts

Direct Answer Uncertainty propagation carries variability from exposures, hazard projections, vulnerability, damage functions, financial forecasts, and parameter estimates into final risk outputs. Hindcasting tests whether a model, using only information available at an earlier date, reproduces subsequently observed conditions or outcomes. Together, the methods address different questions: propagation quantifies conditional uncertainty, and hindcasting evaluates historical predictive performance. Validation uses time-aware data splits, preserves real-time model versions when possible, and...

read more

Model Risk, Uncertainty, and Validation: Combining Scenario Ensembles and Alternative Models

Direct Answer Scenario ensembles combine results across climate pathways, general circulation models, hazard models, damage functions, or econometric specifications. A simple average is appropriate only when members are comparable and the weighting rule is defensible. Scenarios are conditional narratives rather than automatically calibrated probabilities; do not create expected values by assigning arbitrary probabilities without disclosure. Analysts can report ranges, robust rankings, scenario-specific decisions, or model-averaged estimates when weights are evidence-based. Dependence among...

read more

Model Risk, Uncertainty, and Validation: Decomposing Portfolio Risk and Financial Performance

Direct Answer Portfolio decomposition attributes total climate risk or climate-adjusted performance to assets, hazards, locations, and model components. For additive expected loss, contribution is straightforward: each asset's weighted loss sums to the portfolio total. For nonlinear measures such as tail loss, probability of negative growth, or diversification benefit, marginal and component contributions depend on the selected risk functional. Analysts need to reconcile contributions to the portfolio measure, distinguish standalone from marginal risk, and avoid interpreting a large...

read more