Insurtech & Insurance Economics

Nobel Prize-winning paper introducing asymmetric information theory. Though focused on used cars, it explains why 'bad risks' can drive out good risks in insurance markets, establishing the intellectual foundation for adverse selection analysis.

The most-cited paper in insurance economics. Demonstrates how insurers design contracts to induce self-selection by risk types and proves that pooling equilibria cannot survive competition. Nobel Prize work.

The founding paper of health economics. Analyzes the special characteristics of medical care markets and introduces moral hazard as a fundamental problem in health insurance.

Introduced the conditional correlation test now standard across the field for detecting asymmetric information. Using French auto insurance data, found no evidence for asymmetric information in that market.

Showed that private information operates on multiple dimensions—risk AND preferences—explaining puzzling cases where riskier individuals don't always buy more coverage.

Introduced the graphical demand-and-cost-curve approach that became the empirical workhorse for welfare analysis in insurance markets. Provides a unified framework for estimating welfare loss from inefficient pricing.

The RAND Health Insurance Experiment—gold-standard randomized evidence showing price elasticity of healthcare demand around -0.2, meaning cost-sharing reduces utilization by 25-30%.

Foundational paper for understanding risk-sharing and reinsurance. Introduced conditions for Pareto-optimal risk allocation and the concept of companies trading risk as a commodity.

Comprehensive review synthesizing ~100 articles on ML in property and casualty insurance. Documents the evolution from GLMs to gradient boosted machines and neural networks.

Demonstrates the mathematical connections between traditional actuarial methods and deep learning, enabling practitioners to blend actuarial intuition with ML flexibility. Introduces the CANN approach.

Introduced RNN architectures for joint prediction of paid and outstanding claims. Uses gated recurrent units to process loss triangles, improving on traditional stochastic reserving methods.

Highly-cited analysis of 84 studies through Porter's value chain lens. Identifies four major industry tasks: enhancing customer experience, improving processes, offering new products, and preparing for cross-industry competition.