Rising Tides
Simulating the Impact of Risk Rating 2.0 on the Distribution of Dynamic Flood Risk in Houston, Texas
P. Mueller (TU Delft - Technology, Policy and Management)
S. Balakrishnan – Mentor (TU Delft - Technology, Policy and Management)
Amineh Ghorbani – Mentor (TU Delft - Technology, Policy and Management)
David Bresch – Mentor
Lukas Riedel – Mentor
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
Climate change is increasing coastal flood risk by synchronizing once-independent flood losses into correlated, systemic events. Under this pressure, the U.S. National Flood Insurance Program (NFIP) has accumulated a structural deficit exceeding twenty billion dollars. In response, FEMA is introducing Risk Rating 2.0 (RR2), the largest pricing reform in the program’s history. RR2 replaces zone-based premiums with property-level actuarial pricing, turning insurance premiums into a property-level risk signal. By elevating their homes, households can reduce both flood risk and insurance premiums, potentially creating a feedback loop between adaptation and affordability. However, adaptation requires substantial upfront investment. This thesis therefore asks: what is the long-term impact of Risk Rating 2.0 on the distribution of dynamic flood risk in Houston, Texas?
To address this question, a coupled human–flood agent-based model (ABM) was developed for the Houston–Galveston Bay area, representing 43,294 residential properties over a 30-year period. The model replicates both the legacy NFIP premium methodology and the RR2 pricing methodology as paired treatment and control scenarios, exposing identical households to identical flood hazards so that differences are attributable solely to the pricing regime. Household decisions on home elevation and insurance coverage are modelled using cumulative prospect theory under evolving risk perception and budget constraints. The analysis evaluates physical exposure, financial exposure, and adaptive capacity following the IPCC AR6 risk framework to examine the premium–adaptation–risk feedback mechanism introduced by RR2.
The results show that the RR2 risk signal functions as intended in encouraging adaptation. Households increasingly substitute physical risk reduction for insurance coverage, resulting in substantially higher home elevations but lower insurance penetration. Expected annual flood damage decreases modestly, while the reduction in insurance coverage increases uninsured structural losses and widens the protection gap. Although flood risk declines across most census tracts, no evidence is found for the emergence of local risk hotspots. Instead, the aggregate results conceal an increasingly unequal distribution of risk. The risk signal is most effective for middle-income households with relatively high flood exposure and lower property values, which can afford adaptation and benefit most from premium reductions. Lower-income households are often unable to finance elevation measures, while wealthier households face relatively weak incentives to substitute insurance with adaptation. As a result, both physical and financial risk shift toward more vulnerable households. Under a changing climate, this redistribution becomes stronger because the system no longer reaches a stable equilibrium. Consequently, RR2 not only responds to climate risk but also becomes a source of social vulnerability by transferring risk from the insurance pool to individual households.
Methodologically, this thesis presents the first coupled human–flood ABM to incorporate FEMA’s complete RR2 pricing methodology together with an endogenous premium–adaptation–risk feedback loop. The model combines empirically informed parcel-level data with heterogeneous household behaviour and cumulative prospect theory and is fully documented and open sourced for future research.
The findings have important policy implications. Reporting expected annual damage alone conceals who ultimately bears flood losses and therefore masks the widening protection gap. Rather than subsidizing insurance premiums, policy should prioritize means-tested support for home elevation, enabling vulnerable households to benefit from the adaptation pathway created by RR2. Although the model does not account for household relocation, changing behavioural preferences, or fully dynamic climate projections, it provides valuable ex-ante insight into the long-term distributional consequences of risk-based flood insurance. More broadly, the results are relevant beyond the United States, as many countries are considering similar risk-based insurance systems under increasing climate risk.
Files
File under embargo until 17-07-2027