The Revolving Door of Risk: Climate Hazards, Risk Containment, and the Hidden Social Dynamics of Managed Retreat

WHAT IS MANAGED RETREAT ACTUALLY MANAGING?

In the United States, FEMA’s home buyout initiative is not the only government program encouraging managed retreat, but it is the largest by far (Mach et al. 2019). Started in 1993 mainly to assist rural farming communities, the program has since expanded into metropolitan areas, where total expenditures to date have reached nearly $4 billion to purchase and demolish approximately 45,000 housing units nationwide (Horn 2025; Elliott et al. 2020). According to official framings, what is being managed by these expenditures is broadly twofold. For participating homeowners, rising flood risks and impacts threaten property values. The buyout program manages this threat by offering a fair, pre-disaster price for a given property, even if that property has been recently damaged (see Siders 2022; Siders and Gerber-Chavez 2021). For FEMA, the threat being managed is rising costs to its National Flood Insurance Program (NFIP), which has been financially insolvent for decades. In 2017, Congress canceled more than $16 billion in debt so that the program could continue covering claims for that year alone. The program’s debt has since ballooned again to more than $20 billion, with the devastating hurricane season of 2024 expected to dig that financial hole even deeper (FEMA 2022).

To accomplish its aims, FEMA also manages how its buyout program works. First, it mandates that flooded areas do not automatically receive associated funds, even after a federally declared disaster; instead, local authorities must apply for those funds. Second, respective applications must include complex cost-benefit calculations showing where and how the requested funds, if awarded, will result in future savings to the NFIP through foregone claims achieved by the proposed purchase and demolition of homes in designated areas. Finally, FEMA requires that all homeowner participation in these areas be voluntary. This provision minimizes political backlash that can arise from the exercise of eminent domain by local officials using federal funds. Notably, there is no requirement to track where participating property owners (or their tenants) move once a deal is struck, or to assess if those moves lead to safer housing. Instead, program evaluation focuses overwhelmingly on property acquisition, demolition, and cost savings.

Behind these official framings, however, lies a larger, unspoken reality: FEMA cannot possibly afford to purchase all homes at elevated flood risk, no matter how large the projected savings to its NFIP. This is partly the government’s own doing. Since the NFIP was established under the Department of Housing and Urban Development in 1968, it has subsidized the construction of more than four million homes within its designated floodplains, which experts now consider to be woefully inadequate as currently mapped (First Street Foundation 2023). Even if the count of four million homes is believable, at an average price of $420,400 nationwide, the aggregate cost to purchase them all would sum to approximately $1.7 trillion—420 times what FEMA has spent on its managed retreat program to date. This realization means that the program is also working to manage something much larger and more symbolic than its stated aims: namely, the government’s legitimacy in the face of climate challenges that continue to grow steadily beyond its control.

In this vein and borrowing from Scott Frickel and James R. Elliott (2018), we propose that managed retreat may be usefully viewed as a policy of “risk containment.” Here, risk containment refers to what federal agencies do politically when they cannot possibly address the full scale and scope of the environmental challenge they face. So, instead they develop practices to contain the perceived risks of the challenge rather than directly engage its fuller material reality. One way that government actors do such risk containment is by focusing their interventions on the most obvious, or visible, sites of concern. This practice redirects public attention away from the outsized scale of the broader environmental challenge and focuses it instead on a few discrete, or contained, sites of acknowledged concern. Here, public attention can then be channeled into smaller, more technical conversations about related rules, procedures and site selection, all while projecting a sense that something larger and more meaningful is being done.

To illuminate this dynamic, Frickel and Elliott (2018) focus their research on the US Environmental Protection Agency’s (EPA) regulation of urban brownfields. Their analyses show that the agency only recognizes approximately 12 percent of all likely polluted sites, at best. We extend this thinking beyond sites polluted by former industrial activities to sites threatened by future flooding, and in the process from the EPA to FEMA as a federal agency of growing environmental significance. In making this extension, we maintain the view that the elements of risk containment that managed retreat represent need not be deliberate or ill-willed. They can arise simply as an unintended consequence of being outmatched financially and programmatically by the daunting environmental challenge flooding and other climate hazards now present. What we do modify, however, is our analytical scope. Instead of spotlighting all sites of concern (in our case, all housing units located within local zones of retreat), we focus only on residents who voluntarily move from these areas following policy implementation. In taking this approach, we deliberately reframe managed retreat as a mode of climate mobility rather than property acquisition.

INFLUENCES OF COMMUNITY ATTAINMENT AND ATTACHMENT ON MOBILITY

As we turn attention from property acquisition to residential mobility, it is vital to do so in sociologically informed ways. The first way, we contend, is by moving away from the imagery of isolated homeowners making atomistic decisions about if and where to move in the face of rising climate risks and related policy interventions. In such situations, potential movers remain embedded in community contexts that influence how they perceive and act on the risks and opportunities they experience. This embeddedness occurs because homes are much more than placeless residential units and economic investments; they are also social achievements with place-based status and personal attachments. As we invite these status and attachment considerations back into the conversation, we expand discussion of managed retreat beyond technically rational cost-benefit assessments to engage enduring insights from urban and community sociology. This dialogue, in turn, can help illuminate what is familiar and what is new about today’s climate mobilities. To assist, we begin with classic insights from the Chicago School. We then follow those through subsequent modifications and cultural critiques. We then end with a Du Boisian caution about the scientific pitfalls of applying related insights in colorblind ways. The overarching aim is to begin peeling back the community dimensions of residential relocation to better see the hidden social dynamics of climate mobilities, particularly in zones of active retreat.

The first systematic efforts to theorize community dimensions of residential mobility came from the early Chicago School of urban ecology. While notably blind to the power of racism (more on this topic in the next section), the school advanced the idea that even seemingly static, place-based communities are often quite dynamic demographically. This dynamism occurs as established residents leave and newcomers succeed them, resulting in a steady churn of households through local areas over time (Burgess 1925). Considering how difficult it was (and remains) to observe such residential churning beneath the surface of net population statistics, this insight was quite sophisticated for its time. Yet, Ernst Burgess (1925) and colleagues’ use of analogies from plant ecology to motivate and explain this residential “metabolism” struck many scholars as misguided and overly deterministic. In response, subsequent researchers did not abandon the idea so much as substitute economic reasoning for its motivation, introducing the concept of “filtering” along the way (Temkin and Rohe 1996).

Traceable to Homer Hoyt (1933) and later Wallace Smith (1963), filtering assumes the presence of locally stratified housing markets produced by an ongoing interaction of economic forces and institutional actors. The latter include landlords, developers, and realty boards who profit from treating the basic stuff of place, the land, as a commodity, thereby investing variably in its construction, marketing, and upkeep (Besbris et al. 2024). As different local submarkets form and shift over time in relation to one another, households are assumed to strive to move into successively better, or more valued, submarkets over their life course. Here, better can mean different things depending on one’s life stage: a better school when raising children; a shorter commute when balancing career and home; a nice, waterfront property when looking for one’s forever home. But, beneath these varying tastes is presumed to be a common cultural expectation to pair spatial mobility with social mobility, that is, to move up, not just out, when relocating voluntarily. This expectation is supported materially when one’s housing generates wealth that can be subsequently reinvested in new housing in more affluent residential areas over time. As these individual- and community-level dynamics entwine, households filter upward into successively higher-valued communities, opening housing vacancies in their wake. These vacancies, in turn, are filled by others following behind them, reinforcing community attainment as a powerful social dynamic shaping residential mobility.

This filtering perspective, while still influential, met cultural critiques from the beginning, which continue to this day (see Firey 1945; Gans 1962; Small 2004). A common complaint is not so much that aspirations and behaviors of community attainment do not exist. Instead, it is that by emphasizing such forces, researchers miss important social and cultural influences that generate deep and lasting attachments to place-based community, and thus residential immobility. These influences include neighborly bonds, networks, institutions, and sentiments that hold people voluntarily and affectionately in place. Evidence of such attachment has been uncovered time and again across a wide array of cases and places, including prominent communities fighting against urban renewal and gentrification (Korver-Glenn and Mayorga 2024) as well as those recovering from recent wildfires (Morales-Ginder and Mook 2025) and flooding (Lynn 2017). In the latter, the power of community attachment often shows up through interviews with stayers, that is, households that after a recent environmental catastrophe decide to remain in place rather than relocate elsewhere (see Rhodes and Besbris 2022; Kimbro 2022). In telling their stories, field researchers often forefront the difficult emotional, social, and material work that residential immobility now requires in the face of rising climate risks and repetitive impacts.

As debates over community attainment versus community attachment persist, we contend that pitting the two social forces against one another, as if diametrically opposed, is increasingly misguided in the face of today’s rising climate challenges. Instead, we propose synthesizing the two perspectives to foreground how both community attainment and community attachment intersect to structure climate mobilities, both managed and unmanaged. In this vein, we propose that climate movers generally strive to satisfice both social dynamics by seeking not only to move up socioeconomically when they retreat but also stay geographically proximate in ways that help retain their community attachments, even while relocating. We conceptualize this conjunctural force as one of attached attainment. Further, we propose that, if households cannot actualize it, they will be reluctant to relocate, deciding instead to stay in place. In this way, attached attainment influences not only where people move as they retreat but whether they do so at all.

RACIAL PRIVILEGE AND THE REVOLVING DOOR OF RISK

As W. E. B. Du Bois (1899) pointed out more than a century ago, the canonical theories and debates excavated and synthesized earlier came largely from White scholars studying White communities in ways that failed to adequately account for racism in community-mobility dynamics. This failure has since manifested in numerous ways, including a tendency to assume that what happens in White communities generalizes to Black, Asian, Latino, and other socially marginalized communities of color. Research has long since demonstrated that this is not the case. As a result of oppression and opportunity hoarding that favors White communities, non-White communities have long lacked access to the same political and economic resources that have historically facilitated the upward filtering and protected community attachments found in many White residential spaces. Managed retreat offers no exception. Nationwide, analyses show that more than 80 percent of homes purchased through FEMA’s buyout program have been in majority-White census tracts, offering homeowners in these areas an extra, publicly funded tool to adapt residentially to rising climate risks that threaten their housing as both investment and home (Elliott et al. 2020).

For purposes of the present study, these enduring Du Boisian insights have two important implications. First, they warn against assuming the propositions developed thus far hold beyond the majority-White communities from which they originated and managed retreat now concentrates. Thus, we restrict our ensuing empirical analyses to majority-White zones of retreat. In doing so, the intent is not to devalue investigations of climate mobilities in other types of communities. Rather, it is to illuminate how such mobilities are unfolding in areas where racial privilege in housing access, wealth accumulation, and policy implementation are conjoining to steer where and how managed retreat as a federal policy is now largely operating. A second implication of Du Bois’s (1903) insights draws more directly from his work on “the veil,” which elucidates how White racial privilege can bring with it notable blind spots. In the case of managed retreat, these blind spots can include a relative inability among those filtering up into racially privileged zones of retreat to see the rising flood risks they are taking on as they move into their new homes.

To the extent these dynamics prevail, they begin to conjure an image that looks less like a one-way exit from areas of growing threat—a common conceit in popular accounts of climate migration—and more like a revolving door of risk that continues to spin slowly in place. This occurs as White privilege and risk containment work in concert to maintain steady flows of newcomers, even as nearby housing is being demolished by federally funded local authorities. In this way, displacement via managed retreat and serial resettlement of nearby housing in the same area are not mutually exclusive forms of climate relocation, even on the front lines of retreat. Instead, they co-occur. In Soaking the Middle Class, Anna Rhodes and Max Besbris (2022) illuminate this dynamic in their documentation of why residents first moved into the majority-White suburb of Friendswood, despite its high flood risk. As the authors explain, “moving to Friendswood is an act of social reproduction whereby families are leveraging their existing economic and racial advantages to buy into a place that will yield more advantages in the future” (Rhodes and Besbris 2022, 27). And, when this future brings some of the worst flooding in US history, as Hurricane Harvey did in 2017, newcomer demand continued. When asked why, new arrivals commonly cite quality schools, personal safety, and “the Friendswood feel.”

These sentiments are also evident in the recent ethnography In Too Deep. Here, sociologist Rachel Tolbert Kimbro (2022) recounts the stories of mothers striving to recover in their White, middle-class community following three major floods over a three-year span. Despite real costs to their mental, marital, and physical health, residents explained to reporters that, “dealing with the risk of hurricanes and flooding is part of life, a tradeoff for a neighborhood with sought-after schools, a strong sense of community, spacious lots and a close-in location” (Luck and Cowen 2022). And, as in Friendswood, market demand in the area continues to reflect this collective sensibility. Even after repeatedly flooding, the desire of newcomers to move into the area has increased housing prices by more than 38 percent, assisted by public investments in the construction of a new $23 million elementary school, $50 million cultural center, and $480 million flood mitigation project—all as federal buyouts occur in the same area. Similar trends are also apparent nationwide (Harris 2025).

SUMMARY AND NEW CLIMATE SIGNALS AT DESTINATION

Drawing from the preceding discussion, we advance three hypotheses for empirical investigation, all scoped to majority-White communities, where racial privilege and managed retreat policy intersect with enduring expectations and resources for both community attainment and attachment. First, consistent with recent conceptualizations of risk containment, we hypothesize that market-based resettlement—or business as usual—dwarfs managed retreat as the dominant mode of climate mobility, even in areas where the latter is federally funded, publicly touted, and visibly conducted. Second, we hypothesize that all households moving from these zones of retreat—regardless of their mode of exit—will exhibit behavior consistent with both community attainment and community attachment, meaning they will strive to move up socioeconomically but stay close spatially in ways that integrate these two enduring community influences on residential mobility; otherwise, they will tend to stay in place.

To these hypotheses, we add a third. Namely, that climate signals are now inserting themselves into these enduring social forces to pull climate movers into less risky housing as they filter up into nearby neighborhoods of higher value. In this way, materially speaking, retreat is currently unfolding more as an act of privilege than duress, even within the nation’s most at-risk places.

DATA

Our analyses investigate zones of retreat nationwide. We define these zones by identifying owner-occupant participants in FEMA’s Home Mitigation Grant Program (FEMA 2024) and then drawing a half-mile buffer around their acquired property. To locate these homes and identify respective homeowners by name, we use a national database that was publicly released following a suit filed under the Freedom of Information Act (Benincasa 2019). A map showing where these buyouts occurred, scaled to the number of transactions, appears in figure 1.

• Download figure
• Open in new tab
• Download PowerPoint

Figure 1.

Areas Where FEMA Has Funded Managed Retreat, by Number of Participants

Source: Elliott and Wang (2023b). Reprinted through Creative Commons license.

A stylized depiction of a local zone of retreat appears in figure 2. We use a radial buffer rather than administrative units (say, census tracts) to define our zones because boundaries for the latter may include bodies of water that flood into adjacent units unrecognized if no buyout occurred there. We use a half-mile distance because that length is short enough for nearby residents to know about buyouts in their zone. It is also long enough to account for residents who may not have received buyout offers but nonetheless considered moving in response to local flooding and implementation of managed retreat. We focus only on buyout locations where the program participant was an owner-occupant because we want to ensure that those voluntarily taking observed buyouts are also the ones moving.

• Download figure
• Open in new tab
• Download PowerPoint

Figure 2.

Zones of Retreat Defined Using a Half-Mile Buffer Around Each Buyout Address

Source: Authors’ rendition.

RESULTS

We start by assessing the extent to which residential relocation from zones of retreat occurs through the local managed retreat program versus the market once the former commences. We then assess patterns of attached attainment and address-level risk scores at destination.

Policy Versus Market Moves from Zones of Retreat

To begin, figure 3 presents two ways of viewing the counts of each type of mover over time. In both panels, the x-axis refers to the year in which buyouts began in respective zones. For example, values for the year 2007 refer to the counts of movers from zones where buyouts commenced in that year, not the number of residents who moved in 2007. Figure 3, panel A then displays the national counts of both types of movers on a single y-axis. Here, we see that market movers exceed policy movers by a wide margin. Over the full span of our study, the ratio is approximately 12:1, and that ratio remains relatively constant year to year. At an average count of 9 policy movers per zone of retreat, this ratio equates to approximately 108 additional market movers that typically go unaccounted for in research focused on managed retreat participants. To assess the extent to which zones of retreat with higher shares of renters are driving the larger counts of market movers, we ran supplemental regression analyses. Results (available on request) show no statistically significant association between local proportions of renter-occupied housing and the ratio of policy movers to market movers in the zone of retreat.

• Download figure
• Open in new tab
• Download PowerPoint

Figure 3.

Counts of Movers from Majority-White Zones of Retreat over Time, by Type of Move

Source: Original analyses by authors based on data described earlier.

Next, figure 3, panel B uses a dual y-axis to help visualize the ratio of policy movers to market movers over time. Overall, the results show the two types of movers generally track each other over the course of our study, suggesting both types of movers are responding to the same impacts and risks in their zone, just at vastly different scales. This is true even in later years, as both types of mobility decline. In zones where buyouts commenced in 2016, for example, we count 34 policy movers nationally compared with 614 market movers in the same zones of retreat. The notable exception is 2013—the year in which FEMA began funding large numbers of buyouts in coastal New Jersey following Superstorm Sandy. We are uncertain why this anomaly occurs.¹

Patterns of Attached Attainment in Relocation

To assess if movers exhibit patterns of community attachment and community attainment, or attached attainment, we begin with community attachment. The presumption here is that shorter moves reflect stronger attachment to the home and community from which movers are relocating. Using driving distance (in miles) between origin and destination addresses, we find the median distance for all movers is just 7.4 miles. This is considerably shorter than the median distance of 10 to 15 miles for all moves observed by the National Association Realtors (NAR 2024) over the past thirty years. It is also substantially shorter than the average American’s commute from home to work, which studies now estimate to be approximately 13 miles (Bricka et al. 2024). Thus, it seems most movers from zones of retreat are remaining notably attached to their communities of origin rather than moving away to far-off “climate havens” being promoted by local boosters farther north and inland (Morris et al. 2023).

To visualize this pattern, figure 4 restricts our sample to just those who moved within 50 driving miles of their home of origin (77 percent of all movers in our study). Figure 4, panel A presents a box-and-whiskers plot for this subsample, subdivided by mover type. Here, stripped of outliers, the median distance (represented by the box’s midline) drops to 4.8 miles, with a range of 1.9 to 10.9 miles between the 25th and 75th percentiles, respectively (represented by the box’s lower and upper bounds). Figure 4, panel B then displays a kernel density plot for distances moved for the same subsample. Again, we see the bulk of moves occurring over relatively short distances, as well as similar distributions for policy and market movers.

• Download figure
• Open in new tab
• Download PowerPoint

Figure 4.

Distance Moved from Origin to Destination Address (Driving Miles)

Source: Original analyses by authors based on data described earlier.

Next, we turn to community attainment. Our hypothesis, again, is that movers seek not only to remain close as they retreat but also filter up socioeconomically into higher valued neighborhoods. To test this idea, we begin with the average change in median housing value from origin to destination tract for all movers. The result is a positive increase of approximately $33,000, or 15 percent of the average housing value at origin. This evidence of upward filtering accounts for nearly 70 percent of all movers in our study.

To probe further, we estimate an Ordinary Least Squares (OLS) regression equation predicting change in median housing value from origin to destination for all movers. We keep the model simple, consistent with Stanley Lieberson’s (1985) call to avoid overly complicated models in pursuit of social explanations of complex phenomena. First, to assess variation between policy movers and market movers, we include a binary indicator distinguishing the two. Second, because the flood risk of one’s home may impact the ability to filter up, especially through the market, we include this risk factor at origin as both an independent variable and moderating factor. Finally, to control for changes in the potential to filter up over time, we include a continuous variable measuring the years elapsed since 2007, the first year of relocation observed in our study. Results, estimated with robust errors to account for the clustering of movers from the same buyout tract of origin, appear in table 1 and are then graphed in figure 5.

• Download figure
• Open in new tab
• Download PowerPoint

Figure 5.

Predicted Difference in Median Housing Value Between Origin and Destination Tracts, All Else Equal

Source: Original analyses by authors based on data described earlier and presented in table 1.

Note: All other variables held constant at sample means.

View this table:

• View inline
• View popup
• Download powerpoint

Table 1.

Ordinary Least Squares Regression Results Predicting Difference in Median Housing Value Between Origin and Destination Tracts (Robust Standard Errors)

Broadly, two insights emerge. First, as figure 5 shows, the expected upward mobility of market movers consistently exceeds that of policy movers. This finding conforms to the idea that managed retreat typically serves as a backstop for movers when market prices begin to soften, thus offering less profit to invest in subsequent community upgrading. Second and relatedly, figure 5 shows that as a home’s flood risk increases, subsequent attainment—though still notable—declines for market movers. This, too, makes sense, if one imagines that such risk cuts into market valuation and thus profits that can be reinvested into subsequent community attainment. To affirm the robustness of these findings, we also conducted supplemental analyses that introduced fixed effects for state of origin. Results (available on request) were in the same direction and even stronger than those reported in table 1 and displayed in figure 5.

Climate Signals as One Retreats

Finally, we assess whether people moving from zones of retreat appear to be processing climate signals not just at origin but also at destination. To begin, we compute the average flood factor at each origin and destination address for all movers. For policy movers, we find the mean value at origin is 5.7; at destination it declines to 2.0, or by 65 percent. For market movers, the scores are 2.8 and 2.2, respectively, reflecting similarly low values at destination despite comparatively lower flood risk at origin. These findings are consistent with the idea that the riskiest homes in zones of retreat are the ones most likely to enter the buyout program. They are also consistent with the idea that, regardless of one’s initial level of risk and mode of exit, movers from zones of retreat are seeking out safe homes as they relocate.

To probe further, we reran the same OLS regression model above (in table 1) but changed the outcome variable to the mover’s flood factor at destination. Results appear in table 2 and are graphed in figure 6. Here, again, we see clear evidence of risk reduction for both types of movers at nearly all origin risk levels, even after controlling for other factors. In supplemental analyses (not shown), results affirm that this risk reduction is also clear and statistically significant after the introduction of state-level fixed effects. Most movers, it appears, are looking not only to engage in attached attainment but also to ensure their next location has relatively low risk of future flooding.

• Download figure
• Open in new tab
• Download PowerPoint

Figure 6.

Predicted Flood Factor at Destination Address, by Flood Factor at Origin, All Else Equal

Source: Authors’ tabulation based on data described earlier and results in table 2.

Note: All other variables held constant at sample means.

View this table:

• View inline
• View popup
• Download powerpoint

Table 2.

Ordinary Least Squares Regression Equation Predicting Flood Factor at Destination Address (Robust Standard Errors)

DISCUSSION

A growing literature predicts that more and more Americans will begin moving away from housing threatened by rising climate threats, especially flooding. To help guide these developments, local authorities are partnering with federal agencies to implement a policy of managed retreat. In studying the largest of these programs, run by FEMA, we advocated against thinking about the climate mobilities that emerge as entirely new phenomena. Instead, we encouraged viewing them as being forged through the enduring and often hidden social dynamics of community attachment and community attainment as they intersect with persistent racial inequities and new climate signals. Our results support this perspective. They show that in racially privileged White communities, where managed retreat has thus far predominated in the United States, residents are indeed departing in notable numbers. Our findings also reveal that most of these moves are undertaken by residents who can afford to filter up into safer housing located in more affluent neighborhoods nearby, often through conventional market means rather than through publicly funded property acquisitions. These findings matter for several reasons.

First, they indicate that nationally, our current phase of managed retreat remains a relatively privileged affair. Not only is it disproportionately occurring in majority-White communities, but within these communities, it is also disproportionately occurring among those who can afford to move up socioeconomically while they stay close spatially as they transition into safer housing. Hitting this trifecta is attractive because it offers a way to maintain the social value of one’s community by staying geographically proximate and increasing the economic value of one’s housing investment by moving into a more desired neighborhood while also reducing risk to that investment by ensuring one’s new home has relatively low risk of future flooding. Meanwhile, residents unable or unwilling to make such privileged moves and newcomers entering behind those who leave comprise an increasingly segmented population. One segment is likely to have the financial resources needed to defend in place, say, by paying for additional insurance or elevating their home, which now costs approximately $75 per square foot, on average. Another segment is unlikely to afford such investments and will instead face rising insurance costs that provide declining coverage as they await the next flood in their zone of retreat. Joining them is likely to be incoming residents who, as they replace those leaving, are more motivated by the opportunity to move up into their new community than capable of immediately investing in costly flood protections for their new home.

Although we did not study them directly, we suspect the more socially vulnerable segments of these populations will be notably larger in flood-prone communities of color than in the majority-White communities we did study. This supposition rests partly on the ongoing devaluation of housing in racially minoritized communities in general (Howell and Korver-Glenn 2021), which can block opportunities for “upward exit” to safer housing as well as home equity loans that can help finance new defenses in place. As a result, strong place attachments will likely prevail over new attainment opportunities, leaving residents with little choice but to resist buyouts and instead advocate for place-based flood infrastructure that has long been underfunded in their politically marginalized communities. In Houston, for example, local officials initiated a program to buy out more than 600 homes in a historically flood-prone Black community. Fewer than a dozen residents accepted (Lynn 2017). Faced with this resistance, local authorities began experimenting with alternative ways to encourage relocation from flood-prone communities of color. One way they settled on was to shift such efforts away from FEMA funding toward other sources of revenue that allow for the exercise of eminent domain. This strategy is now being pursued in low-lying, predominantly Hispanic communities, where many residents—lacking safe, affordable housing nearby—are reluctant to relocate despite repetitive flood impacts to their homes and neighborhood (Bonnyman 2024).

Another implication of our study involves how retreat is occurring when it does happen. Our findings show that in respective zones, it resembles less of a one-way exodus of property and people than a revolving door of risk, powered by ongoing demand for housing. As this door spins, it does not eliminate local flood risk; instead, it simply transfers it to the next incumbents. A key question thus becomes, how long can this highly individualistic mode of adaptation persist? While the answer remains unclear, what does seem certain is that the federal government cannot afford to acquire and demolish all homes at growing flood risk. Thus, for financial reasons alone, the federal government is likely to continue operating managed retreat more as a form of risk containment than as a comprehensive means of addressing the full scale and scope of the nation’s rising flood challenge. Another thing that seems certain from our results is that no matter how rational relocation from respective zones may seem, residents’ decisions about if and how it occurs will continue to filter through powerful social forces that condition who leaves and to where.

How managed retreat can be refashioned to better and more equitably intervene in these matters remains an issue of ongoing urgency, as does consideration of other mechanisms highlighted by studies in this special issue. Such mechanisms include rising rents in disaster zones, as highlighted by Brian Y. An and colleagues (2026, this issue), which can increase residential turnover in and around zones of retreat in ways that continue to spin the revolving door of risk highlighted in the present study. Another mechanism, raised by Megan Mullin (2026, this issue) in her piece, is the propensity of local officials to support buyouts more for low-income than high-income residents, which not only raises concerns of equity but can also lead to unexpected and increasingly draconian results if low-income communities are reluctant or unable to comply. More conceptually, efforts to improve managed retreat policy can also benefit from engaging more sociologically informed understandings of real estate, which are now synthesizing insights from Du Bois and Karl Polanyi to better illuminate the structural forces shaping questions of why some environmentally vulnerable people and places seem to gain while others continue to lose in ongoing efforts to adapt to our growing climate crisis (Besbris and Robinson et al. 2024).

CONCLUSION

The present study conducted the first nationwide investigation of residential relocation from areas of managed retreat. Analyses focus on majority-White communities, where analyses show that most movers are still relocating through the market rather than through publicly funded programs that demolish at-risk housing on exit. This dynamic creates a revolving door of risk, even in areas where government officials are dutifully seeking to remove not only people but property from harm’s way. Findings also indicate that those “on the move” from these zones of retreat are mostly residents who can afford to engage in attached attainment as they relocate to wealthier, safer housing nearby.

As future research engages these findings, it is worth noting some limitations of the present study. One limitation is that our analyses remained relatively simple by design. We chose this approach because current theorization of climate mobilities is still developing, and we preferred to generate new insights to assist that effort rather than to test more nuanced assumptions with more complicated models. Another limitation is that we did not study flood-prone communities of color. As Du Bois’s methodological work underscored more than a century ago, engaging these communities will take more than simply collecting additional data and making statistical comparisons across different community types. While such efforts can be useful, it is also imperative to think deeply about how race, space, and place intersect with rising climate risks not just to “trap” certain populations in places of increasing risk but also to shape how they understand what is possible and right in response.

Relatedly, our study is limited in what it can say about growing concerns over climate gentrification, defined here as the unintended displacement of less-privileged residents from areas with lower climate risk (Keenan et al. 2018). As the prospect of these more involuntary forms of climate mobility increases, they expand into questions about how different types of climate risk, say from wildfires, may modify and extend managed retreat policy into new areas (McConnell and Koslov 2024). We look forward to continuing to contribute to these important, ongoing efforts alongside others in this special issue.