This research examines how property-level flood risk information, specifically First Street’s “Flood Factor”, reshapes the US housing market when disclosed on major platforms such as Redfin and Realtor.com. The study identifies significant re-pricing of risk and a subsequent redistribution of property ownership based on financial constraints.

Introduction

Flood-related natural disasters have caused devastating losses across the United States, with annual damages estimated at hundreds of billions of dollars. Historically, homebuyers relied on coarse and often outdated Federal Emergency Management Agency (FEMA) maps to understand flood hazards. The public release of property-level, forward-looking flood risk data via major online listing platforms—which account for over 85% of online traffic—marks a significant shift in information accessibility. This research quantifies the price and distributional impacts of this disclosure using a difference-in-regression-discontinuity (diff-in-RD) design, exploiting internal cut-offs in First Street’s risk categories.

Data and summary statistics

The authors assembled a new nationwide transaction-level dataset spanning 2016 to 2024, covering more than 25 million housing transactions. This was combined with property characteristics from CoreLogic, borrower demographics from the Home Mortgage Disclosure Act (HMDA), and voter registration records from L2. While over 80% of properties in the sample have a minimal risk score of 1, a significant fraction faces material risk. Specifically, 1.84% and 0.80% of sold homes were labelled with “extreme” Flood Factor scores of 9 and 10, respectively, indicating a high likelihood of flooding over a 30-year horizon.

Price effects

The research identifies economically meaningful price effects resulting from disclosure, particularly for the riskiest properties. Homes disclosed as having “extreme” flood risk (a Flood Factor of 9) experienced a significant 3.3% price discount relative to those labelled as “severe” (a Flood Factor of 8). These findings are robust across different specifications, including those using higher-order polynomials and various auto-selected bandwidths. The concentration of negative price effects amongst the most at-risk properties suggests that the qualitative labels provided by housing platforms serve as a powerful signal to prospective buyers during the early stages of their search.

Heterogeneous effects

Geography plays a crucial role in the market’s response to flood risk disclosure. For coastal properties, the price decline is substantially larger, reaching 11% on average. Conversely, non-coastal or inland homes show little price response despite having similar cost exposure. This evidence is consistent with a theory of inattention, where flood risk is physically more salient in coastal markets located near bodies of water. Notably, the strongest effects—a 16% average discount—were found for coastal properties situated outside of FEMA-designated 100-year floodplains, suggesting that areas without robust flood insurance requirements are where First Street’s data provided the largest update to consumer perceptions.

Additional validation and robustness

To ensure the validity of the diff-in-RD design, the authors conducted several diagnostic tests. They confirmed the absence of “bunching” at the risk cut-offs, indicating that market participants cannot manipulate underlying risk scores. Covariate balance checks showed no statistically significant discontinuities in dwelling attributes, such as lot size or building age, across thresholds. Furthermore, placebo tests for the pre-disclosure period (2016–2019) revealed that treatment effects were statistically indistinguishable from zero before the information became widely available on housing platforms, ruling out differential pre-trends.

Days on market

Beyond transaction prices, disclosure also affected the duration of listings. Homes labelled as having “extreme” risk stayed on the market for approximately 7 additional days compared to those with “severe” risk. For coastal properties, this increase was even more pronounced, with houses remaining on the market for roughly 13 additional days on average. This evidence is consistent with a reduction in demand for high-risk properties following the public release of granular, property-level risk data.

Distributional effects

The research documents a striking redistribution of risk by household income following disclosure. Buyers of homes just above the “extreme” risk cut-off had 5.3% lower income and were 1.8 percentage points more likely to utilise Federal Housing Administration (FHA) financing. This resorting was most pronounced for coastal properties, where the income difference reached 13%. Additionally, buyers of riskier homes were found to be 1.4 years older on average. These findings suggest that while high-risk homes have become cheaper, they are increasingly being purchased by households that are more financially constrained.

Household resorting through a discrete choice model

A structural mixed-logit model was employed to decompose the drivers of resorting into “price” and “taste” channels. The analysis revealed that the redistribution of flood risk to lower-income households is primarily driven by price changes rather than heterogeneous preferences for risk. Specifically, price changes account for 91.9% of the resorting effect. Lower-income households do not necessarily have a higher tolerance for flood risk; rather, they are more price-sensitive and thus more willing to trade off risk for the relatively cheaper prices of high-risk properties once disclosure has successfully incorporated that risk into market prices.

Conclusion

The disclosure of property-level flood risk has led to a significant re-pricing of environmental risk, particularly in coastal areas. While the provision of better information improves allocative efficiency, it also raises critical questions regarding climate justice. As high-risk homes become relatively cheaper, they attract financially constrained buyers who are least able to afford the long-term costs of potential flood damage. The results suggest that information disclosure alone may exacerbate existing inequalities in exposure to environmental hazards as risk falls on the shoulders of those least able to pay to avoid it.