Risk from extreme precipitation and climate change on New Zealand’s residential property
Three manuscripts form the basis of this dissertation exploring the effect of extreme precipitation and climate change on residential property in New Zealand. The first manuscript investigates the public insurer’s expected future liabilities, given future climate projections. Specifically, it examines the effect of extreme precipitation on direct property damage associated with rainfall-induced landslides, storms and floods. This study applies a fixed-effects panel regression model using claim data linked to extreme precipitation data over 2000-2017 and future climate change scenarios until 2100. The results show that liabilities will increase more if future greenhouse gasses emissions are higher. At the aggregate level, the percent change between past and future liabilities ranges between an increase of 7 to 8% higher in the next 20 years, and an increase between 9 to 25% increase by the end of the century, depending on the greenhouse gases emissions scenario.
The second manuscript examines the risk of property damage from landslides associated with extreme precipitation. The focus is on the Nelson region as it displays the highest number of claims and pay-outs relative to its population and residential stock asset, and two thirds of the pay-outs come from a single event. The focus is on this event. This research combines past insurance claim data with geographic and sociodemographic data to estimate probability of damage, which is then combined with property replacement values and damage-ratio information to calculate the expected loses and map the spatial distribution of risk. The study integrates into the risk estimates the impact of climate change on precipitation based on an ‘attribution’ study. The analysis shows that slope and social deprivation play a significant role in the probability of damage. Furthermore, higher expected losses are associated with higher property values.
The third manuscript studies the current and future risk of property damage from floods associated with extreme precipitation and climate change. The focus is on the most expensive event on record. This study applies a logistic cross-sectional regression model that exploits spatial variation of rainfall intensity-duration-frequency (with and without the effect of climate change), while controlling for other factors that might make a property more or less likely to experience damage. The expected monetary losses are calculated by factoring in the likelihood of flood damage derived from the regression model, property replacement values, and property vulnerability (based on flood-depth fragility functions). The results show that highest losses are associated with lowest annual exceedance probabilities (AEPs), still, sizeable losses are associated with higher AEPs. In this case, the effect of climate change for different emissions scenarios is too small to cause an economically meaningful increase in risk levels in the next 80 years (2100).