This dissertation consists of three chapters. The first chapter investigates the impact of insurance affordability criteria on uninsured losses. The vehicle for analysis is a model of a catastrophe insurance market with explicit representation of the key stakeholders (homeowners, primary insurers and reinsurers). The theoretical framework is applied in a case study of eastern North Carolina with spatially explicit representation of hurricane damage due to wind and storm surge. We evaluate the effect of an affordability threshold expressed as a percentage of a home value. If the cost of insurance exceeds 1% or 2% of home value the home is uninsured due to the affordability constraint. We find that the homes that fail the affordability test account for a high proportion of expected losses in the high risk region or our study area and subsidization of insurance rates would not be cost effective. The second chapter investigates the impact of insurance affordability criteria on uninsured natural catastrophe losses and addresses the question of whether a voluntary, affordable catastrophe insurance market is viable. We use the same game theoretic modeling framework and the same study case used in the first paper. Examining affordability thresholds of 1% or 2% of home value, we find that homes that fail the affordability test account for a high proportion of expected losses in the high risk region of our study area, public subsidization of insurance rates would not be cost-effective, and private subsidization would destroy incentives for insurers to participate in the market. We conclude that a combination of insurance, retrofit, and acquisition is necessary to address regional hurricane risk. The third chapter develops a dynamic modeling framework for the natural catastrophe insurance market. This framework includes (1) discrete choice models for the homeowner insurance purchase decision; (2) discrete choice models for retrofit; (3) explicit modeling of the interaction between competing insurance carriers and a heterogeneous homeowner population; and (4) a probabilistic representation of hurricane occurrences over time.