(Mis)trusting the process: how post-disaster home buyout processes can degrade public trust

Federally funded housing buyout programs are the dominant method of government-supported retreat in the USA. Done correctly, buyouts can reduce pre-disaster vulnerability and facilitate post-disaster recovery. However, the success of buyout programs hinges on successful coordination and implementation by local administrators, who represent buyout participants, manage the buyout process at the community level, and connect them to state and federal resources. Because of this, trust between local administrators and the members of their communities is crucial for project participation and successful outcomes. While local administrators play a critical role in the buyout program, their role in building trust throughout the process has been an understudied aspect of the buyout literature. To address this gap, our paper examines the perceptions of local buyout administrators related to trust. This is done through a study of the conditions following Hurricane Matthew’s landfall in North Carolina, USA, in 2016 using in-depth interviews with 18 local HMGP administrators, and an analysis of over 300 local newspaper articles to study how trust is built and lost in the buyout process. Our findings indicate that a lack of program clarity, unclear communication about the program’s guidelines across all levels of governments, and extended time frames deteriorated public trust in a manner that hindered program success and diminished program results.


Introduction
After a presidentially declared disaster, the US federal government provides aid and assistance to disaster-affected communities through partnerships with states. One part of the recovery process may include relocating households out of risky areas, which allows for both future disaster mitigation for communities and household-level recovery for residents (Binder et al. 2020). This is typically undertaken through the provision of federal funding through "buyout" programs, the largest of which is the Federal Emergency Management Agency's (FEMA) Hazard Mitigation Grant Program (HMGP). This program provides funding to local jurisdictions through state-level governments to purchase significantly damaged or high-risk homes in the aftermath of disaster declarations from voluntary homeowner participants. Through HMGP, FEMA has helped fund the buyouts of close to 40,000 properties over the last three decades (Mach et al. 2019a, b). While largely considered a cost-saving success (Rose et al. 2007), households are often wary of participating because the buyout process can be lengthy, bureaucratically cumbersome, and lack transparency (Binder et al. 2015).
Scholars and practitioners argue that social trust between households and buyout administrators is a key ingredient to household participation in HMGP, and shapes perceptions and experiences of overall project success (de Vries and Fraser 2007;Perry and Lindell 1997;Saunders and Smith 2020). However, there is scant research in understanding how social trust is gained or lost through the buyout process (Greer et al. 2019). Further, there are limited studies examining the role of local administrators in the success or failure of buyout programs (Sipe and Vella 2014;Smith et al. 2021). To address this gap, our paper examines the role of trust in administering HMGP at the local level, with a focus on the following research questions: 1. What features of the buyout process deteriorate public trust? 2. How are local administrators affected by the relationships developed with the public through past HMGP programs, and how might current participation shape future programs? 3. What is the role of local administrators and government in building or eroding trust in the buyout process?
To answer these questions, this study analyzes in-depth semi-structured interviews of local administrators who oversaw buyouts in eastern North Carolina following Hurricane Matthew in 2016. We used semi-structured interviews as they suited for situations where open-ended questions are necessary and follow-up queries are anticipated (Adams 2015), and the intent is to explore the thoughts and beliefs of the interview subject (DeJonckheere and Vaughn 2019). In addition, we gathered over 300 newspaper articles related to Hurricane Matthew buyouts from North Carolinian newspapers to understand the disaster recovery context over three years and the public's evolving perceptions of the buyout process. Newspaper analysis of disaster coverage has a longstanding history in research (see, for example, Friedman and Wenger 1986;Wenger and Quarantelli 1989) and can reflect efforts to frame the event (Rausch 2013). Using mixed methods was done to add analytic richness (Fielding 2012) and to triangulate the data to add scope and depth by including information from different time periods and perspectives (Flick 2009).
Our analysis suggests that the program's protracted timeline and the prevalence of misinformation deteriorates local trust, while emotional support and labor by local administrators helps to build and protect trusting relationships. Furthermore, our work looks at how past HMGP processes may affect the ongoing HMGP implementation in the aftermath of Matthew. Additionally, we discuss how the post-Matthew recovery process colors perceptions of the ongoing HMGP buyouts for Hurricane Florence, which struck many of the same areas and towns in North Carolina in 2018. These findings 1 3 offer insights into the importance of protecting and building trust in the disaster recovery process broadly, and buyout programs specifically.

HMGP participation and trust
After presidentially declared disasters, different funding streams become available to state and local governments, as well as affected residents. Buyouts funded through the HMGP requires an arduous process, and multiple levels of governmental cooperation (Kihslinger et al. 2017). Through HMGP, households can receive support for post-disaster mitigation options that include home elevation, home reconstruction, or a buyout (which may also be referred to as relocation, or acquisition projects), where a damaged property is purchased for the pre-flood value, and the household is able to relocate as desired. Connecting residents to federal funding, however, requires support from local and state level agencies (Ji and Lee 2021). Their roles in this process are described briefly below: Local government administrators and officials act as representatives for interested residents. When funding becomes available, local administrators and officials have conversations with interested residents about the appropriateness of HMGP. If the resident decides that they would like to apply for HMGP funding, the local government applies to the appropriate state, territory, or tribal government on their behalf as a "sub-applicant." In the application, the local government must emphasize the perceived long-term benefits of the mitigation effort (such as reducing future flooding) and how the proposed actions align with the states hazard mitigation priorities as defined in the state hazard mitigation plan (Kihslinger et al. 2017).
The state receives these applications and ranks them based on state priorities and requirements at the state and federal level. While the primary eligibility criteria are cost effectiveness, environmental considerations, and technical feasibility, the state may consider other objectives when determining project priorities. For example, they may look to prioritize the properties with the highest benefit-cost analysis, concentrate buyouts within certain high-impact areas, distribute funding across all declared counties, or develop other selection criteria. Because major hurricanes often occur years apart, priorities may not be the same from one HMGP funded recovery to the next (Smith 2014).
The state submits the approved and ranked project to FEMA, who confirms federal eligibility requirements are met. If adequate funding is available, funds are first distributed to the state-level government, and then to the local governments. Once they have the approval, the local government can proceed with HMGP projects, including buyouts. This process, from disaster to approval, takes an average of 1.9 years, after which, it commonly takes several more years for the project to close out (Weber and Moore 2019).
Although the program and process is daunting, HMGP and other buyout initiatives have the potential to reduce community and household level risk, create amenities, and provide taxpayer cost-savings, which can make them broadly appealing (Rose et al. 2007).There are additional benefits when local governments can buy out contiguous parcels, which can be put to more productive use. For example, contiguous parcels can be better transformed into amenities for the town (e.g., creating a park), and can even reduce future flood risk for surrounding properties through adaptive measures (Salvesen et al. 2018;Zavar and Hagelman III 2016). However, because the HMGP requires voluntary participation, 1 3 willing involvement on the part of impacted residents is necessary for the program's success (Binder et al. 2020). Belief that HMGP programs will be run efficiently and effectively is necessary for engaging greater numbers of individuals, which creates more flexibility in prioritizing the most at-risk households, and allows for the formation of contiguous open land. Therefore, creating conditions that encourage more eligible participants to apply has compounding benefits.
The current literature establishes a number of factors that make residents more or less likely to participate in buyout programs, including pre-and post-disaster community conditions, place attachment, the extent of damage, risk perceptions, and certain demographic characteristics, as noted in a synthesis of 89 pieces of literature covering more than a dozen different disasters by Greer et al. (2019). While these factors are well established, Greer et al. (2019) note that trust is an additional factor that may affect relocation and resettlement decision making, and has the potential to be important, but that not enough information was available to support or explain their effect. In fact, only a few studies center on trust in the buyout process.
For the purpose of this analysis, we define trust in line with the work by Kasperson et al. (1992). As the authors note, trust has been defined in a number of different ways, with each definition conceptualizing trust in the context of different processes (e.g., cognitive, behavioral, emotional, etc.). We rely on the specific concept of social trust, which focuses on the dynamics of trust as it relates to social institutions and societal processes, and how individuals and organizations can facilitate (trust) or hinder (mistrust) engagement. Social trust is defined "as a person's expectation that other persons and institutions in a social relationship can be relied upon to act in ways that are competent, predictable, and caring" (Kasperson et al. 1992, 169). This definition was selected as it is established within the risk perception literature, which has strong ties to hazards literature, as described in greater detail below. Within the buyouts literature, de Vries and Fraser (2007; have undertaken much of the research emphasizing the importance of trust. In some of the earliest work on the subject, de Vries and Fraser (2007) implemented a survey of 152 property owners from Greenville and Kinston, North Carolina, participating in the buyout program after Hurricane Floyd (1999), as well as interviews with 12 local government representatives and buyout stakeholders in the same cities. In their latter work on the subject, de Vries and Fraser (2012) expand the survey and interviews to include residents and administrators in Grand Forks, North Dakota, and San Antonio, Texas, in addition to Kinston and Greenville. In both studies, they found evidence to suggest that mistrust in public employees or in the process made the decision to participate in buyout programs less desirable for residents, suppressing participation. However, while these pieces of literature emphasized the role of trust, the process of how public facing entities might build trust was largely outside of their scope and was only touched upon briefly. We hope to build on this by focusing on the process of building or losing trust, within an established framework detailed further below.

Building and degrading trust
While the work in the hazards literature that has introduced the role of trust in the buyout process is critically important, it does not provide a sufficient theoretical frame for this analysis. For that purpose, we rely on the larger body of literature focused on risk perception for further insight into the trust processes.
Risk perception research has strong ties to disaster and hazards research, and these fields can overlap in many areas. However, historically, risk perception literature has focused more on the role of trust in public office as it relates toward faith in regulations, or the acceptance of or objection to the siting of technologically uncertain industries, such as nuclear or chemical facilities (Slovic 1987;Kasperson et al. 1992). Further work has tied the literature on the public's risk analysis of technology and industries to the public's risk analysis of natural hazards. This work has found that trust is particularly important when residents have low knowledge about the hazard and are faced with uncertainty, as is often the case with natural hazards, as trust encourages reliance on the advice of authorities when residents cannot trust their own personal judgment (Wachinger et al. 2013;Paton 2007).
Risk perception literature shows that a fundamental characteristic of trust is that it is delicate. If you conceptualize trust processes as retaining trust, building trust, and degrading trust, then it is well established that the latter is significantly easier to accomplish than the first two. As a result, events and actions that deteriorate trust have an outsized impact on the overall trust relationship compared to those that build or maintain trust (a review of this asymetry is provided by Slovic 1993). This delicacy illustrates the importance of actions that protect existing trust relationships during turbulent times.
Returning to the definition of social trust, our article centers the interviews with the perspective that protecting, building, and degrading trust occurs in four key dimensions related to levels of: competence, reliability, caringness, and predictability (Ozawa 2018; Kasperson et al. 1992). These terms are defined by Kasperson (1992, 170) and echoed by Ozawa (2018, 23) as shown in Table 1.
Importantly, there is not a process for trust building or deterioration that is consistent across all levels of interactions. Actions that increase trust in individual-to-individual interactions may degrade trust between individuals and institutional representatives (Kasperson et al. 1992). For example, a government administrator may profess a lack of knowledge about an issue to a resident. Based on prior studies, this honesty may increase the trust between the two individuals, but could degrade trust in the institution on the part of the resident, who may see this as a lack of competence (Kasperson et al. 1992). Therefore, it is important to note the level of any interaction that is building or degrading trust and the various ways in which that interaction may influence trust on other levels.
Ozawa (2018) operationalizes this concept of social trust in relation to the public adaption of risky technologies, specifically the siting of a nuclear facility on Three Mile Island in Pennsylvania, and the proposed replacement of the open-air Mt. Tabor Reservoir in Table 1 Kasperson's (1992) explanation for how social trust is built or lost Commitment Trust relies on perceptions of uncompromised commitment to a mission or goal, and fulfillment of fiduciary obligations or other social norms Competence Trust is gained only when the individual or institution in a social relationship is judged to be reasonably competent in its actions over time. While expectations may not be violated if these individuals or institutions are occasionally wrong, consistent failures and discoveries of unexpected incompetence and inadequacies can lead to a loss of trust Caring Trust relies on a perception that an individual or institution will act in a way that shows concern for and beneficence to trusting individuals Predictability Trust rests on the fulfillment of expectations and faith. Consistent violations of expectations nearly always result in distrust… However, predictability does not necessarily require consistency of behavior. Complete consistency of behavior would require unchanging actions or beliefs, even in the face of contradictory information, and also more consistency in values and related behavior than most individuals, groups, or institutions possess Portland, Oregon with underground tanks. She shows that public trust in public-facing institutions deteriorated through misinformation, uncaring actions, incompetent responses, and inconsistent actions. Additionally, she argues that this mistrust hindered project development, and limited agencies' ability to respond to accidents and problems. In doing so, she explicitly links distrust to decreased community resilience to crises. This has clear parallels to buyout programs, which are a negotiation between residents and public-facing agencies about alleviating the siting of individuals within harm's way. Its implications of trust on community resiliency further connect these studies and demonstrate that social trust can be a useful lens for analysis of the HMGP programs.

Study area
North Hurricane Matthew made landfall on October 8, 2016. In North Carolina, nearly 35,000 households experienced flood damage and about 5000 experienced "major" or "severe" home damage (Department of Commerce 2017). According to the Department of Commerce (2017), the geographic extent of storm damage was concentrated, with 52 percent of major or severely damaged homes located in just 13 towns. However, while severe flood damage was focused geographically, the overarching extent of areas affected covered almost half the state. As a result, households in 50 counties across eastern North Carolina were declared eligible for FEMA Individual Assistance disaster aid (see Fig. 1).
Hurricane Matthew's flood damage occurred disproportionately in socially vulnerable communities. Between a half to two-thirds of homeowners affected by the flooding, and 86 percent of the affected renters, had low or moderate incomes, defined as a family income lower than 80 percent of the area median income (Department of Commerce 2017). Furthermore, neighborhoods deemed severely impacted had a disproportionate number of lowincome residents, Residents of Color, and households without access to a car (Department of Commerce 2017). This creates the condition where many of those who are eligible to participate in the HMGP buyouts have limited resources to commit to a lengthy process and a program that inherently has significant uncertainty.

Interviews
Between May and September of 2019, we conducted semi-structured interviews of local HMGP administrators throughout eastern North Carolina. While this was three years after Hurricane Matthew made landfall, approvals for buyout projects had only been processed recently and residents were, in many cases, still waiting on relief. In order to select interviewees for this study, we began by identifying all jurisdictions that were declared eligible for FEMA Individual Assistance, as shown in Fig. 1, and identified which jurisdictions participated in the Hurricane Matthew HMGP. This narrowed our list of potential sites down to 34 jurisdictions. Using local government websites to identify the emergency manager and relevant county employees involved in disaster recovery in each jurisdiction, we developed a list of potential administrators to interview.
We contacted individuals on this list via email and phone to inquire about their role in the recovery process and whether they would be available for an interview. Following this initial outreach, we utilized a snowball sampling method to identify the primary administrator of the HMGP in each jurisdiction. In all but two cases, our interviewees had been the administrators of their local Hurricane Matthew program from the outset in 2016. Each jurisdiction included in the study had at least one property approved by FEMA for an acquisition, reconstruction, or elevation through the HMGP. Additionally, all jurisdictions pursued approval for at least one acquisition project. The single jurisdiction with the greatest number of projects had over 100 properties approved for the program.
We interviewed 18 local administrators who came from a variety of different positions in local government and a variety of different jurisdictions. The communities represented included seven counties, and four towns, who managed a total of 249 HMGP projects. The roles and departments of interviewees included: Planning, Community Development, Emergency Management, Fire Department, Disaster Recovery, Jurisdiction Manager, Building Official, and Public Information Liaison. The diversity in the roles and day-to-day duties of the administrators interviewed reflects the diversity in education, skills, and experiences of those responsible for local hazard mitigation.
Interviews lasted between 15 and 60 min and were conducted remotely via telephone. With interviewee approval, each interview was recorded and transcribed verbatim. Interviews focused on the process of administering the Hurricane Matthew HMGP, centering on three main areas: 1. The process of recruiting program applicants and engaging residents in recovery; Individual Assistance designations open up federal funding opportunities for programs that provide support for disaster survivors after a presidentially declared disaster 2. The perceived capacity of, and available resources for, administering the HMGP in that jurisdiction; and 3. The role and perceived effectiveness of, as well as the nature of communication with, representatives responsible for HMGP at the state level.
Though interviewees were asked to focus on their experiences during Hurricane Matthew, we did ask about the local impacts of Hurricane Florence and whether that storm, which struck in 2018, had affected Hurricane Matthew recovery. Beyond this, several interviewees addressed their work on other recovery programs like the Community Block Group Development-Disaster Recovery program. In addition, three respondents indicated they had prior experience with HMGP in the aftermath of Hurricane Floyd (1999), however, it should be noted that FEMA has significantly restructured since then, and many administrative aspects of the program have changed.
In line with the Ground Theory Method (Glasser and Strauss 1967), for our initial analysis of the interview transcripts, we used an open coding approach. An open coding approach can help to identify new insights through an inductive process (Corbin and Strauss 1990;Blair 2015). We used this process to identify thematic codes that were prevalent throughout all the interviews (Williams and Moser 2019). We then used axial coding to further refine and align the themes into more specific categories (Williams and Moser 2019). Through close reading of the transcriptions, the research team worked collaboratively to develop a codebook and identify relevant keywords. This included six overarching themes: program clarity and efficiency, government coordination, institutional knowledge and capacity, community engagement, timing, and references to other storms, derived from 13 codes each with 1-5 keywords to support coder identification.
Two members of the research team then independently coded each interview using the codebook with MAXQDA software (VERBI Software 2019). We selected individual short paragraphs as the standard unit of analysis, as they better represented coherent ideas. To ensure the quality of the data analysis, Cohen's Kappa (1960) statistic, which can be a more reliable measure than simple percent agreement calculations, was calculated as a measure of intercoder reliability. The Kappa value of 0.65 suggests there was substantial agreement (McHugh 2012) between the independent coders. Once intercoder reliability had been established, we identified the major findings of the interviews by merging the coded data, finalizing the codebook, and manually addressing any outstanding disagreements. At the end of this process, the overarching selective theme that emerged related to trust processes, and the data were interpreted using Kasperson's (1992) explanation of social trust as a framework for understanding.

Article data collection
In addition to the interviews, we also pulled newspaper articles relating to Hurricane Matthew and the HMGP program from local and state-level publications in North Carolina. We used these data to provide additional richness and triangulate the data from the interviews with information from a larger time span and differing perspectives (Fielding 2012). We used the "America's News" database by NewsBank inc., which offers tens of millions of full-text articles nationwide, including documentation from over 800 newspapers. This gave us the ability to search through multiple sources for related articles. Our search was undertaken shortly after we finished coding the interviews, and the information was pulled specifically on 02/10/2020. In this search, we used a content analysis approach to systematically make thematic inferences from large text-based data (Stemler 2001). To this end, we used the following search terms: • "Hurricane Matthew" in 'All Text' AND "Buyout" in 'All Text' • Restricted sources to 'Newspaper' • Restricted location to 'North Carolina' After our initial search, we had 453 hits. Thirty-one of the articles were deleted because they focused narrowly on pig farms (which faced major environmental issues related to Hurricane Matthew, but which were largely unrelated to our focus on the housing buyout program). An additional ninety-three were deleted because they were duplicates or deemed by the research team as irrelevant or failing to capture information on the buyout experience (e.g., a notice that there would be an upcoming community meeting). This left us with 329 articles, which we analyzed for information on hurricanes that took place both before Hurricane Matthew (most notably, Hurricane Floyd in 1999) and after (most notably, Hurricane Florence in 2018). Additionally, we performed text searches for important keywords such as frustration, wait, trust, and relief, as well derivatives (e.g., waiting). Relevant sections around keywords were read in-full to contribute to a greater understanding of the conditions in the aftermath of Hurricane Matthew and the residents' interactions with different levels of government through this process.

Results and discussion
In this section, we will review our findings, using information from both the interviews and newspaper articles. For our framework, we will utilize the conceptualization of how trust is built and lost, as established by Kasperson (1992). We paid specific attention to how trust between the local administrators and the public was perceived to have changed, how the trust between the local administrators and other levels of government was perceived to have changed, and the implications that any changes in trust had for the buyout process. We will cover this through four themes we found echoed repeatedly over the interviews: (1) length of time and frustration; (2) transparency and misinformation; (3) emotional support; and (4) past and future storms. Where applicable, we will include quotes from the interviews and the articles that we feel best exemplifies larger trends we noticed through coding the articles, and through engaging directly with our respondents. We will conclude this section with a discussion on limitations and future work, where we will highlight what this study cannot address, and suggest avenues for future research that may better establish the greater generalizability of highlighted patterns.

Length of time & frustration
As we covered in the literature review, the buyout process is arduous, involves multiple layers of communication, and regularly takes several years to complete. Unsurprisingly, a recurring theme found in the data was that the program delays and the lengthy time frame required for program approval and distribution of HMGP funds created problems for both residents and local administrators, leading to feelings of frustration. In fact, ten of the interviewees emphasized residents' frustration with the prolonged buyout process, as exemplified in the following quote: We have just made the first acquisition offers within the last 30 days [close to three years after Hurricane Matthew]. Just how slow these programs are creates frustration for these people. Some have moved out, some have moved back in, some have [left the program]. -Interview This finding is also evidenced in our analysis of newspaper articles, where mentions of frustration became more prevalent as the process dragged on, with increased hits on this keyword around anniversaries of the disaster, the end of the year, and in August or September of 2018, around the time of Hurricane Florence (see Fig. 2). The following quotes illustrate this point: It has been a year since Hurricane Matthew damaged or destroyed at least 200 properties in Kinston and Lenoir County and officials are still waiting on federal funding for the buyout program. Sam Kornegay, Lenoir County emergency planner, said his office has received many calls from people affected by the flooding that followed Matthew in October 2016. "They are asking 'Where we're at?'" Kornegay said. "'How long is it going to take?' We'll know soon and then they will know. We're on FEMA time." -Article, The Free Press, November 2017 [Community development manager, Andrew] DeIonno added, 'It's not hard to understand their frustration. After all, we're talking three years and the money is there. We understand why they are upset when they call or come out here.' -Article, Tarboro Weekly, Dec 2019 Newlyweds [names redacted] were living in the Forest Hills community in 2016 when their new home of just three months was flooded by Matthew... "I think now the feeling is more frustration than anything because we keep being told, 'Oh, it's (buyout)coming. It's coming. Oh, there's this. Oh, there's that.' So we still have nothing to show for all of the talk that we hear." -Article, The Goldsboro News, September 25, 2018 In some cases, this frustration led to anger against local administrators and accusations of impropriety in the buyout process. These conditions appear to deteriorate public trust in Fig. 2 Number of times "Frustration" was mentioned in our sample of articles over time the local administrators while also straining relationships with state employees whom residents clearly believe share responsibility in the blame for delays. This was evidenced in the interviews with two administrators working for small towns: Well, we've [the administrator and residents] had ups and downs. You know, we've become friends. I think that they trust me, but they got so frustrated with the process and they would get angry and they would be upset and they would accuse me of things. They accused me of not caring because they were Black and I'm white, and if they had been white, this process would have taken, would have been done a long time ago. Both of these quotes are interpretations from the local administrators on how the prolonged HMGP process in the aftermath of Hurricane Matthew created problems for their administration of the program, and led, in their view, to decreased interest in participating in the program after the next storm. These quotes are representative of many of the conversations that we had with the other local administrators to whom we had a chance to speak to. In the first quote, we see how the interviewee felt that there was a base relationship involving trust, that was strained and deteriorated under conditions of frustration, leading to accusations that were both personal and professional. The second quote shows how the feeling of unreliability or uncaring (the state wasn't [there] to help them), led to disengagement from other HMGP opportunities.
Reflecting back to Kasperson's definition of trust (1992) and how it can be built or degraded, we see relationships between these observations by local administrators and the identified themes of commitment, competence, caring, and predictability. The first quote referenced caring directly, but the accusation that the local administrator faced of racial bias suggests that the resident also felt concerns about the commitment of the local administration as well. The perceived failures of Matthew's recovery, identified in the second quote, suggests concerns that could relate to commitment, competence, caring, and predictability. This view suggests that the residents believed that the state, as represented to them by the local administrators, failed them in their inability or unwillingness to follow through on what was seen as a commitment to support recovery. Taken together, we see evidence in our interviews that the length of time was related to a degradation of trust between the public and the local officials. In response to the wait, residents were frustrated, stressed, or upset. This may have a causal relationship with attrition in program participation over the length of the buyout process, or that attrition may be caused by other compounding factors.

Transparency and misinformation
Our analysis suggests that another factor that affects trust is misinformation coming from all levels of government. A number of local administrators noted that they received incorrect, misleading, or simply unclear information from federal or state administrators that they perceived as increasing confusion among participants. Our interviewees noted that this not only created a "trickle-down" effect where passing along incorrect information challenged local administrators' relationships with the public they served, but it also deteriorated the trust local governments had in the state and federal partners, causing further problems with administering the program. This is reflected in other contemporary research, where community officials note that they had a number of problems with assistance provided by the state (Smith et al. 2021). In interviews with local officials responsible for the acquisition of hazard-prone properties through national programs, Smith et al. found that local officials felt that the state and federal government were an impediment to their success. Specifically, one official succinctly put that "we get very little help with [resources, data, policy guidance, or training related to buyouts] from the state" (Smith et al. 2021, n.p.).
In our case, one example of confusing information from the state level occurred when approval letters were sent out to residents that stated that their homes met the standards necessary for participation in the program. According to a county-level administrator, residents and some local administrators interpreted the letter as approving them for a program buyout when, in reality, it was merely a confirmation that they met the minimum standards for approval. However, due to funding limitations, not all residents that received the letter would be approved as not all qualified projects could be funded. Noting this issue, one county-level administrator commented on how this process played out: You know you expect to get the correct guidance at the state level. And not throwing them under the bus-a lot of things they were trying to do and they were trying to make it [the HMGP recovery] as efficient as possible. But the standards of FEMA and the federal government have for these programs is really strict depending on what you apply for and there's not one that mimics the other. And so, if you're not really familiar with those federal guidelines as to how that program has to go-you end up telling people things-and not on purpose-but you end up telling people things that are not entirely true…We were giving information to citizens that was given to us from the state and it was just not true. We've had a lot of negative feedback in that aspect. -Interview This is echoed in news reports of resident experiences: They [a buyout applicant household] asked FEMA what they needed to do in between the time the elevation could be done and then getting the house ready. They were told they could go ahead and rebuild and the house would be elevated once it was finished, she said. So we began building, I believe is was March of 2017," she said. "The house was finished in November 2017. But we got a letter in August 2017 saying that there was no more money and that they wouldn't elevate us. -Article, The Goldsboro News, September 2018 To manage their uncertainty about the veracity and clarity of the information, they were receiving, some local government administrators tried to insulate residents from what they felt were confusing state directives. As seen in the quote below, sometimes this meant waiting to pass along information to participants until they could verify directions from the state or ensure that information from the state would not change. In other situations, it meant placing the onus on state representatives to report information that may have created unwarranted optimism about aspects of the recovery, such as the potential for additional funding streams, as one interviewee notes: Well, a representative from the state was here and discussed that [CBDG Disaster Recovery Assistance]. Because we're very careful about telling them about any additional funds because we don't want to get their hopes up. Because we know a lot of times when the state tells us they're going to do something, we never see it. So, we have to be really guarded with that information. We don't want to be blamed for getting anybody's hopes up and then not delivering. So we let the state bring that up, and we didn't say anything else about it. -Interview As such, this breakdown of trust, not only between the residents and government administrators but also between different levels of government, appears to have impeded the flow of information to those who needed it most. Further, delays in information from the state led to situations where the residents blamed their local administrators for unresponsiveness, a characteristic that feeds into perceptions of uncaringness, incompetence, unpredictability or a lack of commitment and therefore degrades public trust. To address this, some administrators tried to be as upfront as possible with the information that they had been given. Four of the interviews highlighted the importance of transparency as a way of protecting against the degradation of trust, reflected in this statement: We created an entire website on our page on the Hazard Mitigation Grant Program, feel free to look at that. All we can do just put a status update, as of August 9th, we've heard nothing back from the state and are still in a holding pattern. The folks [residents] that I talk to… say you guys have been great; I'm not mad at you, just generally frustrated etc... so I think that the people that we communicate with on a weekly or biweekly time frame, they understand. The distrust might not be there [between local administrators and residents], but distrust boils over to the state or federal level absolutely. -Interview Based on the interview data, we find that confusion and miscommunication between different levels of governments has two main effects. First, it creates frustration and mistrust when residents feel they are being misled, corroding the relationship between the publicfacing local administrators and their constituents. Second, if the local administrators do not trust higher levels of government to provide accurate and timely information, then they may be more reticent to pass on information that they do receive. Overall, these findings suggest that conditions of uncertainty degraded trust across all levels of government, and led to conditions where, rightly or wrongly, residents had less information to act upon. This aligns with previous work in this area (Slovic 1993;Kasperson et al. 1992;Ozawa 2018), which found that misinformation degrades trust between institutions and individuals and can hinder cooperation.

Emotional support
Despite the challenges local administrators face throughout the process, and the resulting friction with residents, local administrators appear to have prioritized the care of their constituents. This is reflected in stories of emotional support for, and emotional investment in, residents throughout the process, which, in turn, interviewees believed led to greater faith in the process on the part of those residents. In fact, this emotional support was seen by many administrators as a core aspect of their job. Referring to the perceived successes of the HMGP program, one interviewee-a county administrator who started his position after Matthew struck in 2016-highlighted the value of connecting and supporting their residents: With me just taking over in December [2018], just being able to contact the property owners, being able to talk to them. Some of them have expressed things that are going on with them and I just try to keep, try to keep them thinking about the positive things, that it's almost to the finish line, we're almost there, just trying to get them those assurances. I'm not in their position but I know it cannot be fun. -Interview The language this interviewee used, where he referred to providing assurances ("we're almost there"), caring about their mental perspective ("keep them thinking about the positive things"), and creating connections with them ("just…being able to talk to them… [about] things that are going on") suggests emotional labor to help manage feelings surrounding the program. This reflects a caring relationship between the local administrator and their resident.
Another aspect of this emotional investment is a feeling on the part of local administrators of responsibility to, and protectiveness of, their constituents. Many local administrators argued that they had a duty to protect their citizens against the confusion and delays they believed were coming from the state and federal level, as noted in the prior section. The following statement notes how this was conceptualized in one department: If there's anything I can tell people it's that if we're waiting on something, it's not going to be waiting on me for something. We want to get the job done on behalf of our citizens. As a county government, we're that first line of defense per se. -Interview This quote highlights an emphasis that the local administrator placed on being seen as a caring and committed partner with the resident through this process. Again, both these aspects are tied to trust building processes through the social trust framework.
These findings suggest that these moments of emotional connection (providing assurances), accessibility (finding out what is going on with the residents and engaging with their stories), and protectiveness were strategies that local administrators used to mitigate the deterioration of public trust even as the HMGP program stretched on, burdening residents and government employees alike. Through these forms of caring and investment, local administrators saw themselves as doing what they could to maintain the goodwill of residents, even as other features of the hierarchical process were beyond their control ("if we're waiting on something, it's not going to be waiting on me"). The fact that this emerged as an important theme across a number of interviews suggests that a greater emphasis within the HMGP program needs to be placed on local government's role in emotional support during the buyout process so that resources can be put toward creating the capacity for local administrators to engage with their residents in this way.

Past and future storms
Because North Carolina has participated in a significant number of HMGP buyouts (North Carolina Climate Office 2019), previous and future storms are often in the minds of vulnerable residents. As a result, we examined the data to better understand how public perceptions about the HMGP recovery process are shaped by the buyout processes, and media coverage of the program, that occurred during previous storms. In this case, the most notable example was Hurricane Floyd in 1999. Further, as two major hurricanes have directly struck North Carolina since Hurricane Matthew (Hurricane Florence in 2018 and Hurricane Dorian in 2019), we theorized that the successes and failures of the Hurricane Matthew recovery program may influence perspectives on future storms and their recoveries. This gets to the core of our second research question: How is participation in HMGP influenced by the experiences of past programs and how might it shape future programs?
We find evidence that prior experiences with disaster recovery, whether positive or negative, can shape perceptions about the current post-disaster context. For example, those that lived through Hurricane Floyd in 1999 and could recall the speed of the associated buyout program-the second largest HMGP project at the time-expressed high baseline levels of trust in government following Hurricane Matthew. Historically, HMGP projects take an average 1.9 years to reach the approval phase (Weber and Moore 2019). In contrast, over 1,000 properties were approved for a buyout within just seven months of Hurricane Floyd (FEMA 2000; Weber and Moore 2019). Because this prior buyout program was considered to be successful, it seemed to fuel interest in participating in the Hurricane Matthew buyout program. This response is captured in this comment: I think that a lot of it came from word of the mouth in the community or a personal history after [Hurricane] Floyd. People who were a little more leery of government were saying that we were going to come in and buy your house and make it all better... They didn't participate after Hurricane Floyd, but that one was so successful so when this one [Hurricane Matthew's HMGP program] came up, we had people that did want to participate because they knew it was going to be successful. -Interview This quote exemplifies how, according to the perspective of the local administrator, success in prior programs led to faith in the local government which then led to increased participation in Hurricane Matthew's recovery program. Prior to Hurricane Floyd, the residents were mistrustful (or, in the words of the administrator, leery) of government support. However, that recovery was seen as successful. This would necessitate positive views of the government's commitment to the fulfilling goals of the recovery program and competence in achieving them. As a result, per the perspective of the local administrator, they were able to capitalize on that good will, thereby creating a base relationship for local administrators to work with the public toward involvement in the program.
The notion that individuals draw relationships between past and future storms is further supported by the analysis of local newspaper articles. In a review of local newspaper articles referencing Hurricane Matthew and HMGP, we found that references to Hurricane Fig. 3 Number of times "Floyd" and "Florence" were mentioned in our sample of articles over time

3
Floyd were common in the initial months after Matthew made landfall (see Fig. 3). Overall, a total of 76 articles mention Hurricane Floyd. After Hurricane Florence made landfall (August-September 2018), Florence recovery also became a feature of many Hurricane Matthew articles. In total, Hurricane Florence was mentioned in 85 articles over 16 months (see Fig. 3).
While reflections back to Hurricane Floyd recovery support the idea that past program success can be parlayed into future program success, it is unclear that the opposite holds true. Despite the issues they identified with the Hurricane Matthew recovery process, many local administrators were not concerned that the Matthew program would negatively impact participation in a post-Florence program. In fact, they expected Florence buyouts to run smoothly because of the short time-frame between the back-to-back hurricanes. The local administrators attributed this to increased experience and capacity, as well as increased familiarity with the program among residents, which the interviewees anticipated would transfer to a more knowledgeable, organized, and streamlined HMGP process, as seen in an example below: Now that we've been through the majority of Matthew grant it's been a lot easier during Florence as far as the knowledge that I have with the intake office. I did interviews by myself. They did not just come in and drop off an application. They had to sit and talk with me and go through the application and any duplication of benefits, any insurance money they had received. So I think it actually helped having already gone through it. And it's made the process with Florence a little easier. -Interview In some ways, the events of Matthew and Florence mirror the learning process that took place after Hurricane Fran in 1996, which was subsequently capitalized on in the response to Hurricane Floyd in 1999(see, for example, Fraser et al. 2003Smith 2014). Therefore, while our evidence suggests that successful recoveries increase interest in future recoveries, it does not necessarily follow that future participation will be depressed by experiences of program mismanagement and delays. Rather, back-to-back disasters and recoveries may create conditions that insulate the effects of negative perceptions on public trust. We theorize that this may can occur in two ways. First, accrued institutional knowledge may outweigh any negative perceptions as the short timespan between disasters almost ensures continuity of experience in most local offices, which can improve capacity and response. Second, the stress of multiple storms may make the program more desirable, even if the residents feel uncertain about the trustworthiness of government representatives. The latter point is illustrated in this excerpt from a news article, which shows how the repetition of disasters depressed interest in returning: After Matthew, a handful of residents tried to return to their houses. They stayed away, Currie [Whiteville city manager] said, after Florence. There's really no one living over there now, it's basically an abandoned area of town," Currie said. "It's sort of like pictures you see from Hurricane Katrina of the Lower Ninth Ward over in New Orleans. -Article, Tribune News Service, December 2019 However, the relationship between repeat storms and recovery participation needs further research to better understand the tension between residents' increased need for services provided or organized by local governments, and a potential loss of trust in these same organizations, considering the impact of multiple storm events. This tension is captured in different experiences that the administrators noted in two of the municipalities that were impacted by both storms: I can say it's kind of interesting. One of the neighborhoods in the city that flooded badly during Matthew. We-to my knowledge-didn't get hardly any applications from those folks during Matthew from HMGP. This time around we've received a lot from that neighborhood. -Interview We did the same format with Florence-have done some of the same format. Have done two public meetings, but not intake yet. But have not had near the turnout as hurricane Matthew… [I think this is because] a lot of people applied for [Hurricane] Matthew, and did not get approved. -Interview The first quote suggests that the administrator felt that the residents were responding to increased distress, while the latter quote notes that a failure to access relief from Hurricane Matthew's HMGP program depressed interest in Hurricane Florence's HMGP program.

Limitations and future work
One important note about this study is that while we did try to triangulate the interviews with the context provided by newspaper articles, the findings rely heavily on interviews with local government officials and administrators. Comments that they had about misinformation or unclear information that they received from the state or FEMA are not verified by other parties. We did not interview state or federal government stakeholders and therefore were not able to obtain a bi-directional perspective on trust. Similarly, we did not interview residents regarding their relationships to local administrators. As such, our results largely reflect local administrator's perceptions of the actions and feelings of other program stakeholders. This is a perspective that has been missing from research to-date, but future studies should explore the role of trust in the buyout process from the perspectives of other stakeholders. In doing so, future research will illuminate whether the dynamics noted in our results are evident throughout different levels of government involved in the process.
Additionally, our interview and newspaper article analysis found that the buyout process for Matthew may impact participation in future buyout programs since the buyout process that occurred in the aftermath of Hurricanes Floyd and Fran in the late 1990s influenced participation in the current Matthew buyout. There is support for this in the literature on the relevance of trust in bureaucratic processes (Ozawa 2018; Kasperson et al. 1992;Slovic 1993) and in buyout programs, specifically (de Vries and Fraser 2012;Kick et al. 2011), as well as studies illustrating how the deterioration of trust is associated with snags in the recovery process (Binder et al. 2020). However, longitudinal studies of buyout processes are called for to examine the evolution of trust from one recovery program to the next.

Conclusion & findings
As we move further into an era of climate change and disruption, buyout programs will only become increasingly necessary and wider spread IPCC 2018). While the HMGP supports one of the longest ongoing buyout programs in the world (Mach et al. 2019a(Mach et al. , 2019b, it is not at all clear that the federal government has been learning from past mistakes or successes to create long-term programmatic improvements or refinements (Greer and Binder 2017). This is specifically evident in North Carolina's state government response, as lessons from Hurricane Floyd were not successfully implemented after Hurricane Matthew (Smith 2014). Our work contributes to the growing literature calling for systematic improvements to the program necessary to ensure long-term success for both the participating municipalities and residents (Binder and Greer 2016;Salvesen et al. 2018;Lynn 2017;Howell and Elliott 2019).
To conclude, we return to our initial research questions. Our first questions asked: what features of the buyout process deteriorate public trust? Consistent with existent literature on trust and institutions (Slovic 1993;Kasperson et al. 1992; Ozawa 2018), we show that the length of time, clarity of information, and timely updates affect the relationship between stakeholders in this process. Specifically, our study suggests that prolonging the buyout process erodes trust between residents and buyout administrators.
In addition, perceived misinformation or incomplete information from higher levels of government led to conditions where local government administrators did not trust their state or federal partners. Specifically, a lack of program clarity and clear communication about program guidelines were an oft noted hindrance to success. This breakdown rippled down to residents. Feeling as if their communication with state and federal representatives was lacking, local administrators reported holding back information from residents for fear that the information would be incorrect or change. Not providing timely information about the program may make residents feel as if they are being misled by the local administrators and, consequently, sow distrust. This is reflected in work by de Vries and Fraser, which found that tensions between local governments and federal agencies deteriorated public trust in the larger buyout team (de Vries and Fraser 2007).
Other local administrators took positions where they consistently conveyed their uncertainty to their residents. This was done through regular updates on what they knew from higher levels of government, what they did not know, and what information they were waiting on. This strategy can cause its own problem because residents may lose faith in the knowledge of their officials. However, local administrators who engaged with it felt it aligned with a duty to protect their citizens and allowed for greater connection with their residents. Indeed, these updates could convey a sense of caring, offer predictability, and place the information into the hands of the residents, increasing their autonomy. Therefore, we recommend that regular conversations or updates, even when no additional information is available, is preferable to waiting for concrete information from higher levels of government.
Though the local government can adapt to poor information from higher levels of government, their perception that they do not receive actionable or timely information demonstrates a breakdown of the process that makes project success harder to achieve. To address this, higher levels of government need to improve their communication to retain trust with the local administrators. We recommend that state agencies develop online portals where residents can track in real-time the status of the program and their application. We also recommend that FEMA adjusts the HMGP program to allow for the provision of faster assistance by providing advanced funding to the states and local governments. Investments should be made in getting resources to residents earlier, even if doing so reduces some degree of confirmation and oversight at the federal and state levels. If this is not possible, stopgap money should be made available to participants without counting as a duplication of benefits. Engaging in the HMGP recovery process requires taking on additional fundssuch as continuing to pay for a mortgage on a property that was, quite literally, underwater, as well as rent or a second mortgage on a home suitable for habitation. To decrease inequitable outcomes of the program, more support needs to be made available between application, acceptance, and closeout.
Local governments see HMGP as a way to mitigate long-term risk, but residents see it as a way to achieve personal recovery (Binder et al. 2020). Thus, the slowness of the process can lead to stress and frustration among residents, particularly those with limited resources who cannot wait to find suitable housing. Evidence from other recovery processes (Binder et al. 2020) indicates that the inability or unwillingness to remain in a program that takes years to support personal recovery leads to major attrition, burdening implementing agencies and affecting the likelihood of program success. Our findings suggest that it would be possible to increase faith in the project through an expedited time frame, ensuring more options in future recoveries that could support a more robust mitigation strategy dependent on connecting adjacent or concentrated properties (Salvesen et al. 2018). This can be achieved through pre-event application development. Additionally, if funding for pre-event planning made it possible to provide for additional staff focused on disaster impacts, then it would be easier for knowledge from one recovery to the next could be retained at the local level.
Our second question asked how administrators were affected by the relationships developed with the public through past HMGP programs. On this issue, our results were mixed. We have some evidence that administrators perceived that residents who witnessed the successful implementation of HMGP after Hurricane Floyd had a higher baseline trust in the process and were more interested in participating in HMGP in the aftermath of Hurricane Matthew. However, more research is necessary to further analyze the relationship between past HMGP programs and participation in ongoing programs. Specifically, more research is needed to show how poor experiences and the recurrence interval of storms influence future participation and outcomes. Regardless, our findings suggest that there is a lingering impact of previous disaster recoveries on future disaster recoveries, creating an increased emphasis on maintaining and promoting project success from resident perspectives.
Regardless of the success of past programs, local administrators can leverage that experience into more successful recoveries. Even poorly executed recoveries can provide administrators with institutional knowledge and a greater background of experience to engage with their residents in a thoughtful manner. We recommend that local administrators work with residents who have prior exposure to HMGP to address both positive and negative experiences that the residents may have had or witnessed. That said, major disasters often happen years apart; to ensure that prior recoveries can be leveraged for future success, local governments should have discussions on how to preserve information about their experiences.
Finally, in terms of our third question on the role of local administrators in building or eroding trust in the buyout process, we see patterns that indicate that it would be valuable to increase support for local governments. One way in which local officials and administrators appear to maintain strong trusting relationships with the community is to be invested in emotional care of their residents. This has parallels with de Vries and Fraser's work (2012) which found that buyout managers built trust through engaging with the client's stories and listening to their frustrations. Emotional investments in the well-being of community members can help to mitigate problems with program delays and misinformation. However, buyout administrators at the local level wear many hats and frequently have commitments beyond the administration of the buyout program.
This finding echoes other literature arguing that limited resources and loss of accumulated knowledge on behalf of local governments can cause buyouts to fall well short of optimized benefits of the program (Zavar and Hagelman III 2016). Therefore, we argue that increased support and resources from the federal or state level throughout the process would have compounding benefits at a variety of scales, from household to regional levels. Some areas for increasing support include monetary supplements for local governments, resource sharing, education, idea exchange on adaptive land use management for newly bought out properties, among others. Additional resources for hiring the appropriate number of grant administrators have shown to be helpful in creating successful recoveries (Smith et al. 2013;Smith 2014).
Beyond the period of disaster response and recovery, local administrators should remain in contact with potential buyout participants in their jurisdiction. Flood risk is not a secret, and local administrators could begin conversations with residents within floodplains in non-disaster settings. Discussions about resiliency strategies could both increase local knowledge and support relationships between residents and their local officials that could then be leveraged in the aftermath of disasters toward more productive discussions on recovery.
Government buyout programs are complex and bureaucratically cumbersome, requiring significant assistance from local administrators for program participants throughout the process. As such, the relationships that local government officials and administrators build with their communities and potential buyout participants is critical to the success of the program. HMGP is a voluntary program and, therefore, requires residents to opt-in to accept a government buyout. If no or few residents choose to opt into the program, the government does not achieve its goals of reducing or mitigating future disaster losses. In contrast, increased buy-in for the program can lend itself to scenarios where local governments gain more use from the bought-out parcels, in turn, increasing returns on investments. The key to achieving this is creating and sustaining a program in which residents can trust.