Financial constraints and short-term planning are linked to flood risk adaptation gaps in US cities

The study addresses how US cities adapt to increasing flood risk due to climate change and investigates why a gap persists between current adaptation levels and those warranted by risk exposure. Prior efforts to quantify municipal adaptation have been constrained by data scarcity, with most evidence based on case studies or surveys. The authors introduce a large-sample textual measure of adaptation from audited financial disclosures to examine the prevalence and drivers of an adaptation gap, defined as adaptation lower than predicted by flood risk. Guided by IPCC AR6 enabling conditions, the study evaluates three hypothesized factors linked to adaptation gaps: behavioral beliefs proxied by political affiliation, financial constraints, and governance quality proxied by planning horizon. The research aims to determine whether partisan leadership reduces adaptation, whether limited fiscal resources constrain adaptation investment, and whether shorter planning horizons are associated with under-adaptation, especially in high flood risk contexts.

The paper builds on IPCC AR6, which identifies enabling conditions for adaptation: behavioral and social factors, financial capacity, and governance quality. Prior US research documents partisan divides in climate beliefs and policy preferences, with Republicans generally more skeptical about climate change, potentially implying lower adaptation if beliefs translate into action. Literature also emphasizes the cost of adaptation projects and finds that financial constraints can hinder adaptation investments, with survey evidence citing funding barriers among cities lacking adaptation plans. Governance and planning horizons are highlighted as critical, with short-termism (analogous to corporate myopia under frequent financial reporting) linked to underinvestment in long-term projects. Studies also show that extreme events can focus attention and spur action. The authors position their contribution against work that relies on case studies or socio-economic proxies, by constructing a scalable textual measure and testing associations between adaptation gaps and partisanship, finance, and planning, while replicating findings that climate physical risk is priced in municipal bonds and showing that adaptation can mitigate such premia.

Design: The authors construct a dictionary-based textual measure of city-level flood-risk adaptation actions using financial disclosures (budgets, Annual Comprehensive Financial Reports, and bond prospectuses) from 431 US cities for 2013–2020. Adaptation is categorized as hard (physical infrastructure such as seawalls, levees, drainage systems), soft (nature-based solutions such as beach nourishment, bioswales, mangrove restoration), and general adaptation. The dictionary contains 177 unigrams and bigrams curated from guidance documents (CDP, C40, IPCC), FEMA and White House nature-based solutions guides, academic literature, and iterative reading of city disclosures from high-risk states to capture municipality-specific terminology. Data: 8726 financial documents were hand-collected. For consistency, the main analysis uses 3161 city-year observations with both budgets and ACFRs, and a subsample of 2011 city-years also includes bond prospectuses. First Street Foundation’s 2020 National Flood Risk Assessment provides flood risk (% of properties with substantial flood risk), aggregated from ZIP to city level. Financial and demographics (population, debt, fund balances) come from Muni Atlas. Political affiliation of city leaders (Republican, Democratic, Other) is assembled from OurCampaigns and hand collection. Planning horizon is the number of years in the capital budget outlook, manually extracted from budget documents. County-level beliefs about local officials’ role in addressing climate change are from the 2021 Yale Climate Opinion Survey. State grant availability for adaptation is hand-collected and summarized by the maximum grant size per state. Text processing and measures: For each city-year, the authors count sentences containing dictionary keywords. Main adaptation equals the total number of sentences across general, hard, and soft categories in budgets and ACFRs. Hard and soft adaptation are counted separately. In the bonds subsample, main+bond adds sentences from prospectuses. Counts are winsorized at the 99th percentile. Alternative measures (keyword counts, grouped keywords, scaling by total sentences, limiting repeat appearances) are used in robustness checks. Adaptation gap construction: Expected adaptation is estimated via regression of adaptation on flood risk (percent of properties at risk), log population, log total sentences, with state and year fixed effects; standard errors are clustered by state. The adaptation gap indicator equals 1 when the residual is negative (actual adaptation below expected). Sensitivity analyses require residuals below more stringent thresholds. Determinants of gap: The probability of an adaptation gap is regressed on indicators of political affiliation (Republican leader), financial constraints (unrestricted fund balance to total expenses, and logged total debt per capita), and governance (capital budget outlook in years), with state and year fixed effects. Cross-sectional heterogeneity is examined by splitting the sample by within-state flood risk (above vs below median), local climate opinion (above vs below median share saying local officials should do more), state grant size (above vs below median $65,000), and household income (above vs below median). Event response: Using NOAA NCEI billion-dollar hurricane data, the study estimates whether adaptation increases more in the high-flood-risk quartile within a state after the first major hurricane event, controlling for population, document length, and state-year fixed effects. Validation and robustness: Validation shows adaptation correlates positively with capital improvement and emergency fund expenditures, earns larger discounts in flood insurance community programs, and mitigates the municipal bond flood risk premium documented in prior literature. Placebo text on public safety does not reproduce these patterns. Results are robust to excluding common keywords (drainage, stormwater), alternative scaling, and different keyword-counting rules. Standard-error clustering alternatives and hurricane-affected city exclusions are tested.

- Descriptive metrics: Across 3161 city-years with budgets and ACFRs, the average main adaptation is 19.94 sentences (hard 14.99; soft 1.63). Budgets average 16.32 adaptation sentences; ACFRs 3.64. In the 2011 city-year bonds subsample, main+bond averages 28.69 sentences, indicating bonds add adaptation information. - Topics: Most frequent terms relate to stormwater and drainage, followed by flood-related measures and infrastructure (seawalls, inlets, levees). Budgets emphasize capital improvement projects and program/department roles; ACFRs emphasize funding allocation; bonds describe intended use of funds. - Adaptation vs. flood risk: Adaptation is strongly and positively associated with flood risk after controlling for population, document length, and fixed effects. A 10% increase in flood risk is associated with a 5.2% increase in main adaptation in 2013, rising to 8.7% in 2020. Flood risk correlates more with soft adaptation than hard adaptation early in the sample, with both strengthening over time. Following first major hurricane exposure, cities in the top within-state quartile of flood risk increase main adaptation by about 19% relative to lower-risk cities. - Prevalence of adaptation gaps: Despite the positive correlation, more than half of cities where over 10% of properties are at risk have main adaptation below that of the average city with less than 10% flood risk. - Political affiliation: Republican-led cities are not more likely to have a main or hard adaptation gap. The soft adaptation gap is about 9% more likely with a Republican city leader, suggesting potential differences in nature-based solutions rather than hard infrastructure. - Financial constraints: A one-standard-deviation lower unrestricted fund balance to total expenses is associated with a 6.6% higher probability of having an adaptation gap (robust for hard adaptation; weaker for soft). For soft adaptation, a one-standard-deviation lower logged total debt per capita is associated with a 7.3% higher probability of a gap. The explanatory power of financial constraints declines over time. - Planning horizon: A one-year shorter capital budget outlook is associated with a 4% higher probability of an adaptation gap. This relationship is persistent over time and robust in hard adaptation; less evident in soft adaptation. - Heterogeneity: Financial constraints matter more in higher flood risk areas and in counties where more residents believe local officials should do more about climate change. State grants reduce but do not eliminate the relationship between constraints and gaps; short planning horizons remain binding regardless of grant size. Variation in average household income does not materially change the relationships. - Validation: Adaptation sentences correlate with greater capital and emergency spending, larger flood insurance-related discounts, and reduced pricing of flood risk in municipal bond spreads, indicating the textual measure captures economically meaningful adaptation activity.

The findings show that US cities generally increase adaptation efforts with higher flood risk and in response to salient hurricane events, but many high-risk cities still under-adapt relative to predicted levels. The lack of a broad partisan association with adaptation gaps suggests that, unlike climate mitigation rhetoric, adaptation actions to tangible flood risks are less polarized, though nature-based approaches may be less pursued under Republican leadership. The results underscore two actionable levers: alleviating financial constraints and extending planning horizons. Limited unrestricted fund balances and constrained debt capacity are associated with under-adaptation, and existing state grants provide only partial relief. Shorter capital budget outlooks consistently correlate with gaps, pointing to governance and planning processes that prioritize short-term considerations. Evidence that adaptation mitigates municipal bond flood risk premia indicates feedback benefits where adaptation can improve financing conditions, potentially creating a virtuous cycle. Together, these findings inform policy discussions on enhancing municipal fiscal capacity and institutionalizing longer-term budgeting to reduce adaptation gaps, especially in high-risk jurisdictions and where public expectations for local action are elevated.

This study introduces a scalable, validated textual measure of municipal flood-risk adaptation and documents widespread adaptation gaps among US cities despite higher adaptation in riskier areas. It identifies financial constraints and short planning horizons as key correlates of under-adaptation, with limited evidence that partisan leadership explains gaps outside of nature-based solutions. The work expands the empirical basis for understanding local adaptation beyond case studies and surveys by leveraging financial disclosures. Policy implications include improving access to funding (e.g., private finance, public–private partnerships, and bond market instruments) and embedding longer-term capital planning to address myopic governance. Future research should establish causal relationships between city characteristics and adaptation gaps, extend the methodology to other climate hazards (heat, drought, wildfire) and geographies, incorporate policy instruments like zoning and building codes, and evaluate the role of green and conventional municipal bonds in financing adaptation.

The textual approach may miss important forms of adaptation not consistently disclosed in financial reports, such as zoning changes, building code updates, retreat or migration policies, and other regulatory measures often contained in separate policy documents. The focus is limited to flood risk and does not capture adaptation to other climate hazards like extreme heat, drought, or wildfire. While associations are robust, the analyses are not causal; unobserved confounders may remain. Measurement relies on the presence and quality of disclosures, which can vary across cities and over time, and certain smaller municipalities are underrepresented due to data availability and reporting formats.