Is flood mitigation funding distributed equitably? Evidence from coastal states in the southeastern United States

The United States Federal Emergency Management Agency (FEMA) provides funding to state and local governments as well as tribes and territories (SLTTs) through its Flood Mitigation Assistance (FMA) grant program to engage in flood risk management efforts. Although all communities are susceptible to flooding, flooding does not impact communities equally. This article contributes to FEMA's goal of addressing equity concerns by examining whether the FMA program is distributed equitably in counties located in eight coastal states in the United States. Using secondary data from OpenFEMA, the Centers for Disease Control and Prevention/Agency for Toxic Substances and Disease Registry, and parcel‐level flood risk data from First Street Foundation from 2016 to 2020, results indicate that socially vulnerable counties are less likely to receive FMA funding, and counties with greater average flood risk are more likely to receive FMA funding. The findings suggest that there is an opportunity for FEMA to improve the FMA program so that funding can be more equitably distributed, such as providing grant writing and application training and support to socially vulnerable communities, educating socially vulnerable communities about the benefits of the FMA program, and extending the application deadline for socially vulnerable communities impacted by flood events.


| INTRODUCTION
In light of rising sea levels along with increased flooding in both coastal and inland areas, communities have been hard pressed to manage and reduce their current and future flood risks (Tyler et al., 2019). The Federal Emergency Management Agency (FEMA) aids communities in these efforts by providing funding to state and local governments as well as tribes and territories (SLTTs) through its Flood Mitigation Assistance (FMA) grant program. Established in 1994, the FMA program offers nearly $175 million in competitive grants to help SLTTs develop plans and implement projects to minimize flood risks (Fraser et al., 2006;Pew Charitable Trusts, 2018). FEMA selects awardees based on applicants' eligibility and the ranking and cost-effectiveness of proposed projects. By providing these grant funds, FEMA aims to reduce or eliminate claims under the National Flood Insurance Program (NFIP) by helping communities engage in activities to mitigate flooding (Carter et al., 2018).
Although all communities are susceptible to flooding, flooding does not impact communities equally (American Flood Coalition, 2020). For example, studies have shown that smaller communities and communities with greater social vulnerability are less likely to compete for federal grant funds (American Flood Coalition, 2020) and participate in federal flood risk management programs like the Community Rating System (CRS) program (Sadiq et al., 2020). "Social vulnerability is a measure of both the sensitivity of a population to natural hazards and its ability to respond to and recover from the impacts of hazards" (Cutter & Finch, 2008, pp. 2301-2306. Additionally, lower income communities are less likely to apply and receive disaster aid due to onerous application processes as well as language and financial barriers (American Flood Coalition, 2020; Domingue & Emrich, 2019).
Recently, FEMA has publicly recognized the need to address equity concerns. For example, FEMA's latest annual planning guidance underscores the agency's interest and commitment to addressing equity issues and to further reduce the complexity of agency processes and procedures. Additionally, FEMA published a Request for Information (RFI), seeking input from the public on how FEMA can achieve its mission of helping people before, during, and after disasters while also advancing equity for all (National Archives, 2021). In this study, equity is treating everyone fairly (Gooden et al., 2009). Relatedly, a recent Government Accountability Office (GAO) report found that between the years of 2012 and 2020, FEMA funding to support flood mapping were significantly lower in areas with higher social vulnerability and underserved communities (GAO, 2021). Together, it is clear that greater attention and efforts are needed to ensure funding to address flood risks are equitably distributed. In short, we situate this article in the context of social vulnerability, because previous studies have demonstrated that socially vulnerable communities are less likely to compete for and receive federal grants as well as participate in federal programs to address flood risks (American Flood Coalition, 2020;Sadiq et al., 2020). Vulnerability is defined as "the susceptibility of social groups to the impacts of hazards, as well as their resiliency, or ability to adequately recover from them" (Cutter & Emrich, 2006)." Hence, in this context, social vulnerability accounts for a community's exposure and ability to cover from disasters.
This article aims to contribute to FEMA's goal of addressing equity concerns by examining whether the FMA grant program is distributed equitably in counties located in eight coastal states in the United States. These eight states with coastlines-Alabama, Florida, Georgia, Louisiana, Mississippi, North Carolina, South Carolina, and Texas-are the focus of this study due to the increasing vulnerabilities and high risk of property damage in coastal areas (Tang et al., 2013). Moreover, 27.1% of U.S. flood fatalities in 2020 occurred in these eight states (National Weather Service, 2021).
This article aims to answer the following two research questions: (1) Is flood mitigation funding distributed equitably across counties in U.S. coastal states? (2) Is flood mitigation funding distributed at higher rates to areas with greater flood risks? Understanding the extent to which the FMA program is equitable has important and practical implications. For example, results will provide evidence on whether FMA grant funds are equitably distributed to areas that are more susceptible to the impacts of flooding. Additionally, there is also the implication in terms of inefficient use of mitigation funding if results indicate flood mitigation funding is not distributed to areas with higher flood risks.
The remainder of this article is organized as follows. The following section discusses the existing literature on flood mitigation funding, social vulnerability, and social equity. The next section describes the data and methods used to determine if flood mitigation funding is distributed equitably, followed by a discussion of the results. This article concludes by discussing the study limitations and recommending ways to ensure flood mitigation funding is distributed equitably in the future.

| Flood mitigation funding
Although there are myriad tools available to governments to address salient public issues (Salamon, 2002), including hazard reduction, efforts to manage flood risks have primarily centered on providing SLTTs grants to invest in mitigation and offering flood insurance to homeowners, renters, and businesses residing in flood-prone areas (Burby, 2001;Cigler, 2017). The three main grant programs available to assist states and localities in managing flood risks include FEMA's Hazard Mitigation Grant Program (HMGP), Building Resilient Infrastructures and Communities (BRIC) grant program, and the FMA grant program (Carter et al., 2018;Fraser et al., 2006).
FEMA's HMGP was created in 1988 and provides funding for mitigation after the Stafford Act has been declared and the president has made a major disaster declaration (Carter et al., 2018;Fraser et al., 2006). The amount of funding SLTTs can receive is between 7.5% and 15% of the total aid FEMA provides to the state for that particular disaster. To receive funding, grantees are required to pay 25% of the costs. The purpose of this grant program is to help communities implement hazard mitigation measures that minimize the loss of life and property from future disasters while simultaneously reducing reliance on future federal response and recovery funds (Carter et al., 2018). This grant program is, thus, not specific to flood disasters. Nonetheless, eligible floodrelated projects communities can engage in under this grant include, but are not limited to, "property acquisition, structure demolition, flood-proofing of structures, structure relocation, structure elevation, mitigation, localized and non-localized flood risk reduction projects" (Carter et al., 2018 p. 17).
Unlike the HMGP, FEMA's BRIC grant program (previously known as the Pre-Disaster Mitigation Grant Program) provides SLTTs with FEMA-approved hazard mitigation plans funds to implement natural and risk reduction activities (FEMA, 2021). In 2021, the BRIC program is prioritizing projects focusing on mitigating risks to public infrastructure and addressing equity concerns among disadvantaged communities (FEMA, 2021). The maximum amount of non-competitive funding a state or territory can receive under the BRIC program is $1 million; there is an additional $919 million of competitive funds available to for mitigation projects across the country (FEMA, 2021).
Finally, FEMA's FMA grant program, which is the focus of the present study, was established in 1994 and provides competitive grants to state and local governments to specifically develop plans and implement projects to minimize flood risks (Fraser et al., 2006;Pew Charitable Trusts, 2018). This grant program is different from the previous two discussed as it specifically provides funding to address flood hazards. FEMA will cover 75%-100% of the costs. The primary purpose of this grant program is to reduce or eliminate claims under the NFIP by engaging in activities that mitigate flood damage to properties (Carter et al., 2018). For fiscal year 2021, the FMA Grant Program has an estimated total funding of $160 million available to SLTTs to engage in flood risk management activities (FEMA, 2021). FEMA selects awardees through an evaluation process that assesses the extent to which the scope of the project benefits NFIP-insured properties and if the project meets select priority scoring criteria (FEMA, 2021). The select priority criteria includes potential measures of flood risk, such as the amount of Repetitive Loss (RL) or Severe Repetitive Loss (SRL) structures, the number of active NFIP policyholders, and participation in the CRS program. FEMA also places priority on projects that benefit areas with higher social vulnerability, measured through the Center for Disease Control and Prevention's (CDC) Social Vulnerability Index (SVI). It is especially important to understand the equitable utilization of the FMA Grant Program. The FMA grant program focuses on properties that participate in the NFIP, and many low-income households do not have flood insurance, the program could be inherently biased toward serving high-income, less vulnerable households. However, it is not clear whether the program utilization is indeed biased. According to Executive Order 13985, Advancing Racial Equity and Support for Underserved Communities Through the Federal Government, all federal agencies, policies, and programs should be approached with the goal of advancing equity for all. This suggests that even if biases are built into the FMA Grant Program toward more wealthy households, FEMA would have an obligation under Executive Order 13985, to advance equity within the program. This is further reiterated by FEMA's Equity Action Plan (2022), which states that among other goals, FEMA is integrating equity to its programs to help "advance equity in their programs".

| Social vulnerability
Social vulnerability is the extent to which people (or groups of people) are at greater risk of harm based on a variety of factors (Singh et al., 2014). Such factors can include income, race, housing status, education level, and so on. Instead of focusing solely on risk exposure to determine risk of harm, social vulnerability includes social systems and power dynamics in the calculation of risk (Bergstrand et al., 2015). Factors determining social vulnerability both form an individual's environment and are formed by the individual's environment (Singh et al., 2014). For example, when an individual is lowincome (one factor included in social vulnerability), they tend not to have health insurance, own a home, and own a car, which contributes to a lower social vulnerability of the community in which they live. At the same time, those who are socially vulnerable typically are limited to living in low-income areas, where high-income employment is sparse, making it difficult to improve their vulnerability, meaning their environment is also contributing to their social vulnerability. There are a variety of different measures of social vulnerability, most of which are based on data from the U.S. census (Rufat et al., 2019). The SVI measure from the Centers for Disease Control and Prevention/Agency for Toxic Substances and Disease Registry (CDC/ATSDR), and the SVI from the University of South Carolina are the two most commonly used measures of social vulnerability in the United States (Rufat et al., 2019). It is important to note that there have been questions raised regarding the validity of these indices, but despite those concerns, social vulnerability indices are still fairly accepted in emergency management and public administration disciplines (Burton et al., 2018;Jackson et al., 2021;Rufat et al., 2019).
The social vulnerability measure used in this study is the SVI from the CDC/ATSDR. The SVI was designed for emergency management and public health communities as a way to identify areas that are prone to environmental hazards (Flanagan et al., 2011). The SVI is composed of 15 indicators (see Figure 1), which are also further grouped into the following four themes: (1) socioeconomic status, (2) household composition & disability, (3) minority status & language, and (4) housing type & transportation. Using U.S. Census Bureau American Community Survey 2014-2018 5-year estimates, SVI ranks U.S. counties according to the 15 indicators and the four themes (Wang et al., 2020). SVI rank varies between 0 (least vulnerable) and 1 (most vulnerable) (Wang et al., 2020). There are limitations with using the SVI to assess the equitable distribution of mitigation grants, which must be acknowledged. Most notably, the CDC suggest that the SVI can be used "to help local officials identify communities that may need support before, during, or after disasters" (CDC, 2022), but was not specifically developed to assess equity and the CDC does not suggest that the measure should be used as an equity measure. However, as previously stated, the SVI has been used in previous studies for this purpose (see e.g., Domingue & Emrich, 2019;Drakes et al., 2021), and the current study adds to this body of research.
Social vulnerability can impact the ability to cope with natural hazards and when individuals are impacted by a disaster, they require external resources to meet their current needs (Drakes et al., 2021). Bergstrand et al. (2015) examined 3126 counties in the United States and found that in general, areas with high social vulnerability had low resilience (as measured through a community resilience index) and vice versa. This suggests that to enhance resilience, social vulnerability should be specifically addressed as areas with high social vulnerability have greater difficulty recovering from disasters (Juntunen, 2004). Hence, such areas need extra support and consideration when recovering from disasters. The authors note that there were instances where the relationship between social vulnerability and resilience either did not exist or was reversed, but there was an overall trend where higher social vulnerability scores were associated with lower community resilience scores. Myers et al. (2008) found that following Hurricanes Harvey and Katrina, areas with higher disadvantaged populations in the U.S. Gulf Coast region experienced more migration out of the areas, which suggests that disadvantaged populations were harder hit by these hurricanes. Zahran et al. (2008) found that among 832 floods in Texas between 1997 and 2001, the odds of casualties the day of the flood was higher in areas with high social vulnerability. Many other factors also impacted the likelihood of experiencing casualties, including whether dams were present and whether the flood was a multi-day flood, but when controlling for these other factors, areas with higher social vulnerability were still more likely to experience injuries or death. In addition to the immediate response to hazards, scholars have found evidence that social vulnerability also impacts the longer-term responses. Indeed, Hamideh and Rongerude (2018) found that in the aftermath of Hurricane Ike in 2008 in Galveston, Texas, social vulnerability influenced whether residents participated in decisions regarding rebuilding. Specifically, the researchers found that many of the low-income residents were displaced from Galveston following Hurricane Ike, which contributed to their limited input in the recovery process.

| Social equity
In emergency management, social equity can be defined as treating everyone fairly when distributing services and resources (Gooden et al., 2009;Guy & McCandless, 2012). More than simply providing equal resources, social equity refers to providing resources to those with the greatest need, a form of distributive justice (Emrich et al., 2020;Gooden et al., 2009). FEMA incorporates equity in its whole community approach to emergency preparedness and mitigation (FEMA, 2020) and has an objective of achieving equitable outcomes for those they serve (FEMA, 2022a(FEMA, , 2022b, meaning resources and response to emergencies should be equitable according to FEMA's standards and goals. In addition, President Biden established the Justice40 Initiative in 2021, which commits to providing "at least 40 percent of the overall benefits from Federal investments in climate and clean energy to disadvantaged communities" (The White House, 2021, p. 1). This initiative specifically identifies 21 federal programs where this should be implemented, one of which is the FMA Program. If communities have sufficient resources, theoretically, they should be resilient.
Because disasters have devastating impacts on socially vulnerable areas, resources are especially needed in such areas. Social vulnerability has been used in past literature to examine equitable distribution of resources (see e.g., Domingue & Emrich, 2019;Drakes et al., 2021). Other measures of examining equitable distribution of resources exist (e.g., the Economic Disadvantage Rural Community Index) this manuscript focuses on social vulnerability to examine equitable distribution of resources.
However, existing research suggests that social vulnerability can impact flood recovery funding. Van Zandt et al. (2012) found that following Hurricane Ike in Galveston, Texas, areas with high social vulnerability had higher rates of applying for FEMA assistance (which is for minor or emergency repairs) and lower rates applying for Small Business Administration (SBA) assistance (which is for major repairs). Van Zandt et al. (2012) did find that areas with a high number of elderly households and a high number of carless households were more likely to receive such funds, but that areas with low social vulnerability were more likely to receive insurance settlements from the storm. Drakes et al. (2021) found that between 2004 and 2018 people of color tended to receive lower amounts of short-term disaster assistance in areas where the need for assistance was high. Interestingly, when examining individual allocations, social vulnerability, when considered as a single measure, was associated with a greater likelihood of receiving assistance, although this relationship was stronger for homeowners than for renters. Finally, studies have shown that there are additional obstacles that prevent communities from participating and receiving hazard mitigation funds or implementing mitigation measures. Miao and Davlasheridze (2022) found that county tax revenue can impact county participation in FEMA's HMGP property buy-out program. The HMGP buys properties in flood prone areas with the goal of reducing building in floodplains, but requires matching funds for participation, suggesting that counties with fewer financial resources through tax revenue could be unable to participate in the program (Maio & Davlasheridze, 2022). Additionally, Petkov (2022) found that locally-financed investment in infrastructure mitigation tends to be lower in communities with high conflict, suggesting that community polarization can be a roadblock for the implementation of mitigation measures. Although not directly related to social vulnerability, this is another potential hindrance to enacting mitigation measures.

| Data and methods
To explore the relationships between social vulnerability and provision of public (federal) disaster mitigation grants, we estimate regression models predicting two different outcome variables: whether a county received approval for FMA funds in the 2016-2020 period, and the total count of FMA funds received over that same period. We expect to find a relationship between social vulnerability and FMA funding, where higher social vulnerability is associated with lower FMA funds. This is because existing research suggests that areas with high social vulnerability tend to be less resilient to disasters, yet receive lower amounts of disaster assistance, thus suggesting inequities (Bergstrand et al., 2015;Drakes et al., 2021).
All models are estimated with robust standard errors. The explanatory variables focus on social vulnerability and the extant flood risk in the county. Because the first dependent variable is a dummy variable to indicate receipt of funding in a particular county, the models are estimated as probit models. The third model uses a poisson regression for count data to model the number of projects in a county, which ranges from 0 to 107 although the mean of 0.6 indicates that most counties received no FMA grants during this period. (See descriptive statistics in Table 1). This approach has advantages of giving the two sets of regression models (i.e., probit and Poisson) identical structure in terms of explanatory variables and thus improves comparability when dependent variables differ. The basic form of the empirical model is: Independent variables SVI, FloodRiskAvg, Flood-RiskSD, and SFHAshare reflect county-level social vulnerability, average property flood risk, standard deviation in property flood risk, and the share of land area in a special flood hazard area (SFHA, an area with a 1% chance of flooding in any given year), respectively. The three flood-risk measures help capture the average degree of flood risk implicitly weighted by properties (for FloodRiskAvg) and by land area (for SFHAshare) while also controlling for the amount of variance in that flood risk within a county. The β parameters thus reflect the conditional association of a marginal increase in one of those factors on the FMA funding likelihood or number of grants. We also include state-level fixed effects (θ) to capture other unexplained state-level heterogeneity in FMA receipt. It must be noted that, while our sample of county data includes all coastal states in the southeastern United States, no counties in Mississippi received FMA funds during this timeframe. Thus, the state fixed effects models perfectly predict the funding outcomes for Mississippi and those observations must be dropped. While, of course, other factors could be controlled for in this sort of specification (e.g., population, income), we purposefully estimate a rather restrictive model so as to avoid overcontrolling for social vulnerability. If a county's social vulnerability is expressed, in part, through its wealth, demographic composition, political affiliations, or other factors, then including these factors in the model could undermine the power of SVI to explain variation in the dependent variable. To paint a richer picture of the role of social vulnerability in predicting FMA funding receipt, we also explore which particular aspects of the SVI might affect the likelihood and count of FMA grants by decomposing the SVI into its main constituent themessocioeconomic status, household composition and disability, minority status and language, and housing type and transportation. The results allow for a more complex relationship between social vulnerability and FMA funding, including the possibility that factors predicting receipt of multiple grants may differ from those predicting just one grant.
We draw our data from the OpenFEMA dataset on the Hazard Mitigation Assistance Mitigated Properties-v2. 1 This dataset contains grant funding from several different FEMA programs but we use data on FMA grant funding only because of its specific focus on projects to eliminate or minimize flood damage. (Unfortunately, the OpenFEMA dataset does not report grant funding dollar amounts for FMA projects.) This project-level data on Hazard Mitigation Assistance Mitigated Properties-v2 lacks personally identifiable information and can only be geographically located at the county or zip-code level. We aggregate FMA grant funding to the county level for all projects in the following coastal states (i.e., Alabama, Florida, Georgia, Louisiana, Mississippi, North Carolina, South Carolina, and Texas) approved from 2016 to 2020. This allows us to match the FMA funding data to the 2018 county-level SVI measure from the CDC/ATSDR. 2 The SVI is generated based on the four vulnerability theme rankings: socioeconomic status, household composition and disability, minority status and language, and housing type and transportation. This SVI measure relies on 2014-2018 data collected by the U.S. Census Bureau as part of its American Community Survey, thus aligning nicely and partly predating the outcome variable of FMA funding received. This is why we selected the period, 2016-2020. SVI values as reported by CDC/ATSDR refer to the percentile ranking of the county, across all U.S. counties with SVI measures, where higher values indicate greater relative social vulnerability. Our focus on social vulnerability-a constellation of many interrelated factors-centers attention on broader contextual concerns for equity rather than narrowly focusing on specific mechanisms, which miss "the forest for the trees" or risk omitted-variable bias in highlighting some mechanisms but not other correlated mechanisms. The multivariate models included two measures of flood risk at the county level. First, is the average of parcel-level flood risk measured by the First Street Foundation's 3 estimate of the parcel being flooded above 15 cm within 30 years. This is akin to a measure of the share of properties in the county expected to be flooded within the next 30 years. Second, we complement this parcel-level average with a land-area-based measure of flood risk: the share of the county's land area that overlaps an SFHA. This is equivalent to the share of land with at least a 26% chance of flooding within the next 30 years. Because of the often high within-county variation in flood risk, we also account for this variance by including the county's standard deviation of parcel-level flood risk measures from First Street Foundation. Table 1 displays the summary statistics for the dependent and independent variables. As already noted, no Mississippi counties had FMA projects approved from 2016 to 2020. The counties in Mississippi, dropped from the main analysis, do not appreciably change the mean values in Table 1. This suggests that, while that state did not have any FMA grant activity, its counties' typical SVI values and flood risks resemble the other coastal states'. Thus, Table 1 shows the mean values for all counties in these coastal states in the rightmost column. Because the number of observations falls to 718 when we drop Mississippi observations for the regressions, Table 1 reports summary statistics for that sample as well. Table 1 shows a great deal of variation in the main explanatory variables (SVI, its components, and flood risk). Yet the great majority of counties did not receive FMA funding, and those receiving funds did not tend to have many funding projects. Specifically, 8% of the 800 counties studied received FMA funding post-2015. Although most counties did not receive any FMA funding, there is substantial variation in the number of grants received by counties. The SVI has a mean of 0.71, indicating that the county average in the coastal states is around the 70th percentile of all U.S. counties in 2018. Though these counties tend to be relatively socially vulnerable, many counties do rank very low in terms of vulnerability. The average parcel flood risk is only 0.11, although some counties exhibit much higher average flood risks. Over 21% of counties' land areas overlap an SFHA on average.

| RESULTS
To show how support for flood mitigation projects varies with social vulnerability, we presents the results of the multivariate regressions (see Table 2). Recall that the dependent variables are whether funds were received after 2015 and the count or number of projects funded during that period. The probit model for the dummy variable indicating whether a county received assistance during the 2016-2020 period essentially estimates the effects of regressors on the probability of receipt. Columns (1) and (2) in Table 2 tell us how those independent variables influence a county's likelihood of being funded. Column (2) shows the results where the SVI is decomposed into its four main themes. Column (3) does likewise except that the dependent variable is the count of projects received. With respect to the flood risk variables, greater flood risk tends to be associated with greater likelihood of receiving FMA grants among coastal states' counties from 2016 to 2020. Greater flood risk does not predict the number of projects funded. For the probit or a poisson model, greater social vulnerability predict lower likelihood of or fewer number of projects funded.
This overarching result should be emphasized. The probit model (column 1) indicates that more social vulnerability predicts lower likelihood of receiving FMA funding. More socially vulnerable counties appear less likely to receive FMA funding, even after accounting for their extant flood risk. This may reflect lower capacity to successfully apply for these grants. By contrast, counties with greater average flood risk are more likely to receive FMA funding. This is true for both average flood risk using the First Street Foundation measure of flood risk as well as the share of a county's land area inside an SFHA. Both average parcel-level flood risk and average land area risk combine to positively predict receipt of FMA funding. Yet while FMA funding appears to allocate toward places with greater flood risk, it tends to flow away from counties with more social vulnerability.
When we extend the analysis to unpack the SVI (columns 2 and 3), we shed light on the ways that social vulnerability operates to attract FMA funding. The SVI component with the strongest, negative effect is the socioeconomic status. Thus, wealthier counties with better employment and education tend to be more likely to receive FMA funds. This effect clearly drives the overall negative coefficient for SVI in column (1). Yet other aspects of social vulnerability also predict receipt of FMA funds, and not in the same direction. Counties with more minority status and language and with more housing and transportation are actually more likely to receive FMA funding. Thus, although FMA funds tend toward counties with less socioeconomic status, they do tend to flow to areas with more minority status and language, and housing and transportation. The poisson model extends this analysis to identify whether these patterns hold for the count of FMA grants received rather than the binary outcome. The results in column (3) show a fairly similar picture-namely strong negative effects of socioeconomic status and positive effects of minority status and language-but also differ somewhat. The count model shows how more household composition and disability is associated with fewer number of projects funded. Also, the positive effect of housing and transportation becomes insignificant in the count model. These associations with FMA funding are conditional upon county-level flood risk, which continues to play a partly significant role in these models. Counties with more land area in SFHAs are more likely to receive FMA funds, though not necessarily a greater number of funded projects.

| DISCUSSION
This study sought to determine (1) if flood mitigation funding is distributed equitably across counties in U.S. coastal states, and (2) if flood mitigation funding is distributed at higher rates to areas with greater flood risks. Results indicate that there is a relationship between social vulnerability and flood mitigation funding, where counties with greater social vulnerability have a lower likelihood of receiving funding. These results indicate that FEMA's goal of equity is not fully reflected in its funding distribution, suggesting that more efforts are needed to ensure equity. This is especially true regarding socioeconomic status. In addition, the results show that counties with greater average flood risk are more likely to receive FMA grants. This result suggests that funding was allocated to counties with higher than average flood risk during the period examined. This finding is in line with the goal of the FMA funds to help reduce or completely eliminate the risk of repetitive flooding and subsequent damage. Previous research is also consistent with this finding, noting that areas with greater loss from floods tend to receive greater FEMA funds to support recovery (Emrich et al., 2020).
The results of this study are in line with previous studies that have found that socially vulnerable areas are disproportionately less resilient to disasters (Bergstrand et al., 2015) and are more likely to receive lower amounts of disaster assistance (Drakes et al., 2021). An explanation for this finding could be that areas with greater social vulnerability are less likely to have the organizational capacity to apply for these grants. This explanation is supported by previous research, which found that communities are not participating in the CRS program due to a lack of resources to put the application together, such as staff, funding, and time (Sadiq et al., 2020). This suggests that FEMA, or other government agencies could offer greater assistance to vulnerable communities in preparing for and writing grants. Comfort et al. (1999) advocates for preparing for disasters before they occur as a T A B L E 2 Flood mitigation assistance analysis. (1) (2) Note: This analysis is for all gulf states except for Mississippi. This is because there is no Mississippi FMA projects observed in the data. Columns (1) and (2) are probit regressions. Column (3) is a poisson regression. Standard errors are in parentheses. *p < 0.1; **p < 0.05; ***p < 0.01. method of enhancing community resilience, especially for areas with high social vulnerability. Providing training to socially vulnerable communities on how to apply for FEMA grants before a flood occurs could be a tool to increase FEMA grant funding once the flood occurs, ultimately making the funding distribution more equitable across communities. Another explanation could be that socially vulnerable communities are unaware of these grants and funding sources, meaning they are less likely to apply for these grants and when they do apply their lack of familiarity makes their grant applications weaker and less likely to be awarded. This is consistent with findings from Van Zandt et al. (2012), who found that following Hurricane Ike in Galveston, Texas, areas with high social vulnerability had lower rates of applying for SBA assistance. Emergency managers regularly work with the media when managing information for crisis response (Boin & McConnell, 2007), and they can incorporate information about FEMA grants following floods in their media strategy, especially in media markets with high social vulnerability.
Finally, another possible explanation for this result is that following flood events communities with high social vulnerability are preoccupied with the challenges posed by the flood events, and do not prioritize or are unable to apply for FEMA funding. Past research suggests that socially vulnerable communities are disproportionately impacted by disasters and are less resilient when confronted with such disasters (Bergstrand et al., 2015;Myers et al., 2008;Zahran et al., 2008). People prioritize their physiological needs, such as food, water, and shelter (Gibson et al., 2012), meaning after a flood a community that experiences significant loss will prioritize physiological needs before addressing secondary needs, such as applying for FEMA funding. Knowing that communities with high social vulnerability tend to be more severely impacted following a disaster, the inequities in FEMA funding following floods could be because communities with high social vulnerabilities are addressing basic needs, but in communities with low social vulnerabilities, basic needs are met, so they can focus on secondary needs such as applying for funds. FEMA can extend deadlines for applying for such funds in communities with high social vulnerability, which could decrease the social inequities observed in this study.
There are a few study limitations worth mentioning. First, the study is about counties in coastal states. Future studies can, in addition, examine inland regions, as they are also susceptible to the impacts of flooding. Relatedly, there may also be a need to examine counties throughout the entire United States to see if the results of this study hold. Second, FEMA uses per capita damage as a metric to guide its decision for mitigation and recovery spending. It is expected that vulnerable communities have fewer and less valuable assets and simply do not qualify for certain programs. Since FMA also targets NFIP properties, it is possible that NFIP penetration is low in low-income and socially more vulnerable communities." Third, the current study does not account for community polarization, which could impact mitigation policies. Past research found that locally-financed investment in infrastructure mitigation tends to be lower in communities with high conflict (Petkov, 2022), suggesting that community polarization can be a roadblock for the implementation of mitigation measures. Fourth, there are other factors that can affect the distribution of FMA funds, such as state-level policies and activities. Future studies should seek to uncover states' decision making processes for utilizing FMA funds. Finally, this study is unable to ascertain the specific reasons why socially vulnerable counties are less likely to receive FMA funding. More research is needed to understand the reasons behind this result. Such a study might entail interviews with floodplain managers or emergency managers in socially vulnerable communities that have been impacted by previous flood events but did not apply for FMA funding. In addition, future studies can also examine whether states with enhanced state mitigation plans have more funded projects than those without enhanced plans. Fifteen states were awarded FEMA approval and understanding whether this impacts the number of funded projects could help guide future mitigation decisions.

| CONCLUSION
The purpose of this study is to investigate whether there is equitable distribution of grant funding under FEMA's FMA grant program. Specifically, the study address two research questions: (1) Is flood mitigation funding distributed equitably across counties in U.S. coastal states?
(2) Is flood mitigation funding distributed at higher rates to areas with greater flood risks? To answer these questions, the study uses data on Hazard Mitigation Assistance Mitigated Properties from OpenFEMA, SVI measures from the Centers for Disease Control and Prevention/Agency for Toxic Substances and Disease Registry, and parcel-level flood risk data from First Street Foundation from 2016 to 2020. The analyses focuses on counties in eight coastal states that are vulnerable to flooding-Alabama, Florida, Georgia, Louisiana, Mississippi, North Carolina, South Carolina, and Texas.
The results indicate that socially vulnerable counties are less likely to receive FMA funding. The fact that no counties in Mississippi-a state with relatively high average social vulnerability (Wang et al., 2020)-received any FMA funding during this time period reinforces this pattern. This finding suggests that there is an opportunity for FEMA to improve the FMA program so that funding can be more equitably distributed. For example, FEMA can provide grant writing and application training and support to socially vulnerable communities, provide targeted marketing to socially vulnerable communities, and extend the application deadline for socially vulnerable communities impacted by flood events. FEMA can use SVI data from the CDC to identify the specific socially vulnerable communities that will be the focus of these interventions. In addition, the results show that communities with higher than average flood risk are more likely to receive FMA funding. This finding suggests that the FMA program may be meeting its stated goal of providing funding to help reduce or eliminate the risk of repetitive flooding. The findings also provide insight for other countries seeking to address equity concerns as it relates to their flood risk management practices. Specifically, other developed countries may consider the extent to which their processes for providing FMA may burden socially vulnerable groups. It is important to note, however, that our results may not generalize well to other countries. More research in non-US contexts is needed to assess the intersection of social vulnerability, risk, and mitigation policies. Together, the results from this study provide initial insights on the characteristics of FMA funding recipients with respect to equity and flood risk, and serves a good foundation for additional studies on this salient topic.

DATA AVAILABILITY STATEMENT
We are able to share part of our data. We will have to request ability to share FirstStreetFoundation data from that organization.