Understanding Household Recovery Following the Colorado Flash Floods (2013)

Publication Date: 2013

Abstract

In September 2013, a heavy precipitation event deposited more than 17 inches of rain along the Colorado Front Range. The resulting floods killed 8 people, destroyed more than 1,850 homes, damaged or destroyed 968 businesses, and caused widespread across 17 counties. Because of the geographic extent of the damage, the Colorado floods are an opportunity to study household and community recovery in widely varying contexts but under the same state and federal recovery regimes. With the support of a Quick Response Grant, we worked to establish a baseline understanding of household recovery in two heavily impacted towns (Lyons and Milliken), with an emphasis on low-income households in the process of replacing affordable housing. Specifically, we asked: 1) what factors influence more successful, or unsuccessful, recovery for disadvantaged households? and 2) what institutions, networks, and/or resources do households access, or attempt to access during recovery? We approached these questions with a mixed methods research design and data gathered through interviews, household surveys, and participant observation in flood recovery activities. In this report, we explain our processes to identify study communities, develop and pilot our household survey instrument, develop a preliminary survey sample, identify additional target populations, and to establish baseline conditions for a longitudinal study of household recovery. The research resulted in several key preliminary findings. At a community level, we found that 1) there is a mismatch between grant programs and community needs; 2) managing recovery related resources and timelines is particularly difficult for small towns; 3) how towns view their role in recovery also varies widely in the region; and 4) that recovery does not take place in a vacuum, and is substantially affected by regional trends and conditions. At a household level our baseline survey finds that many flood-affected households are still in limbo and are struggling with increased expenses and commitments as a result of the flood. We also find that the resources and networks that households rely on for recovery vary significantly across the two communities. We conclude with a few brief remarks about the relevance of our Colorado case study to disaster recovery policies and planning generally.

Introduction

In September 2013, a heavy precipitation event deposited more than 17 inches of rain along the Colorado Front Range, one of the heaviest periods of rainfall in our state’s recorded history. The event triggered flash flooding, landslides, and mudslides across north and central Colorado, from the mountain canyons of Larimer and Boulder County to the agricultural plains to the south and east of the Rocky Mountains. All-told the floods killed 8 people, destroyed more than 1,850 homes, damaged or destroyed 968 businesses, and caused widespread destruction to roadways and other critical infrastructures.

One of the challenging aspects of the Colorado floods was their geographical breadth; seventeen counties were included in the Presidential Disaster Declaration (FEMA-4145-DR), encompassing hundreds of communities and dozens of political jurisdictions. From a research perspective, the Colorado floods are an opportunity to study household and community recovery in widely varying contexts but under the same state and federal recovery regimes. A comparative study of household recovery in different types of communities and in different political jurisdictions would help to identify some of the key variables that contribute to the pace and quality of recovery.

Household recovery from disaster is a long and complex process, difficult to characterize through cross-sectional study or to accurately recall in ex-post interviews. Recognizing these challenges, our study will use a longitudinal research design that interviews participating households multiple times over a several years period, to understand the progress of their recovery and the various factors impacting their options and decisions. The Quick Response Grant provided essential support for the preliminary stage of our study, granting us the resources to identify study communities, develop and pilot our household survey instrument, develop a preliminary survey sample, identify additional target populations, attend meetings and workshops in the study areas, and establish baseline conditions for recovery amongst participating households.

Research Questions

One of the most durable findings in the research literature on disaster recovery is that recovery is an uneven process; disasters tend to exacerbate inequalities in communities, with some households recovering more quickly and completely than others (e.g. Daniels et al. 2006¹; Haas et al., 1977²; Mileti, 1999³; NRC, 2006⁴;Olshanksy, Hopkins, & Johnson, 2012; Wisner et al., 2004)⁵. Whereas early studies of recovery (e.g. Haas, Kates & Bowden 1977; Kates and Pijawka 1977) focused on recovery at the community scale and theorized disaster recovery as “ordered, knowable, and predictable” more recent studies highlight how the terrain of recovery is actually quite varied at the household level (e.g. Rubin 1985⁶; Olshansky, Johnson & Toppings 2006⁷; Ganapati & Ganapati 2009⁸; Rubin 2009⁹; Smith 2012)¹⁰. As Smith and Wenger (2007) argue, recovery is largely a social, not technical, process, and as such unequal recovery is rooted in the same social, economic and institutional processes that reproduce inequality outside of disasters (238; Benner & Pastor 2012).

Our Quick Response study was designed to establish a baseline understanding of household recovery, with a particular focus on low-income and disadvantaged households. Over the course of our longer study we hope to understand several interrelated questions about unequal nature of recovery at the household and community scale:

1. What factors influence more successful, or unsuccessful, recovery for disadvantaged households?

a. What types of assistance do they receive?

b. What types of assistance are needed but not received?

c. Do prior social and civic connections make someone more likely to receive the types of assistance they need?

2. What institutions, networks, and/or resources do households access, or attempt to access during recovery?

a. Do households in higher capacity towns or cities rely on different institutions or networks than those in lower capacity communities?

The data collected with the support of the QR program is vital to the establishment of our longer study and serves as the basis for additional funding applications.[1]¹¹

Methodology

The flash floods caused significant damage across north and central Colorado, with Boulder, Larimer, and Weld counties most affected. To decide which communities we would sample from, we visited several affected communities in those three counties. We also interviewed public officials at the local, county, and state level. These informational interviews, carried out from October-December of 2013, focused on understanding: 1) the variable impact of the floods on different communities; 2) what recovery resources were available to flood affected communities, and the relative time-frames for delivery of outside resources; 3) the relative capacity of different communities to manage recovery; and 4) and recovery priorities for leadership at the local and state level. We conducted approximately 15 interviews, including:

• Elected officials and town staff in Boulder, Estes Park, Evans, Lyons, and Milliken;
• Leadership at local faith-based organizations providing human services;
• Officials at the Colorado Department of Local Affairs (DOLA);
• Officials at Colorado Emergency Management and in the Governor’s Office for Long-Term Recovery and Resilience;
• Officials from the Federal Emergency Management Agency (FEMA), including the Public Assistance (PA) program, the Individual Assistance (IA) program, the Hazard Mitigation Grant Program (HMGP), the Community Planning and Capacity Building Recovery Support Function (CPCB RSF), and the Federal Disaster Recovery Coordinator (FDRC).

We also interacted regularly with volunteer citizens involved in the flood recovery and county staff assisting the towns with housing redevelopment.

Based on these interviews, interactions and associated site visits, we narrowed our focus to two heavily impacted communities, Lyons and Milliken. Lyons is a small mountain town in Boulder County with just over 2,000 residents. Along with Jamestown, Lyons was the community most impacted by the September floods. Lyons sits at the confluence of the North and South St. Vrain Creeks; when both waterways expanded into their floodplains, Lyons suffered significant damage to its natural and built environments. All of the town’s infrastructures were damaged or destroyed, including water sewer, and gas lines, roadways, vehicular and pedestrian bridges, trails, and two major parks. All local businesses were affected, either directly through flood damage and loss of utilities or indirectly through a decreased customer base. The flood damaged or destroyed 215 houses, more than 20% of the town’s housing stock. Affordable housing was particularly affected; the flood damage was concentrated in the Confluence area, where many of the town’s working class population lived. The flood also destroyed both mobile homes parks in town, one of which sat on the town’s broken sewer system, making it unhealthy for its residents to retrieve the belongings that weren’t washed away. Because all of Lyons’ utilities were offline, the entire town was forced to evacuate for a minimum of six weeks. As of February 2014, approximately 145 households were still displaced, representing 15% of the town’s total population, based on a Housing Needs Assessment commissioned by the Town.

Milliken is a fast growing town of 5,610 located in Weld County, Colorado. Milliken is situated at the confluence of the Big Thompson and Little Thompson Rivers. Like Lyons, the floods cut off overland access to the town, making it inaccessible by car or vehicle for the first 48 hours of the disaster. The damage was largely in the northern part of the town, near the creeks and several drainage ditches feeding into them. The major impact was to two mobile home parks, where 35 trailers were destroyed and another 8 were significantly damaged. The town also suffered damage to roadways, some utility infrastructure, and historic structures.

Next, we designed a survey instrument to gather baseline information on flood-affected households including socioeconomic profiles, household work and mobility prior to the flood, social connections and civic involvement, flood impacts, and progress towards physical, economic, and social recovery. The survey includes 70 questions about the respondent and members of their household. The questions are mostly closed-ended, with opportunities for explanation and elaboration. This allows our data to be both quantitative and rich qualitative. After we completed an initial draft of the survey and secured a waiver from the Colorado Multiple Institutional Review Board (COMIRB), we piloted the instrument with five households in Lyons. Based on those experiences, we made minor modifications to the instrument and began the full-survey effort. The surveys are conducted in-person, except in the case of one household who is displaced outside the region. In that case, the survey was conducted over the phone. The average time to complete the survey is 75 minutes, with some lasting as few as 30 minutes and the others lasting 2 hours. As of June 5th, 2014, our total sample size is 27 (see Table 1).

Our total sample is lower than our initial goal of 40 households, for several reasons. The first and most important complication with our preliminary data collection plan was the predominance of Spanish-speaking households in Milliken affected by the flood. We quickly recognized that our English language survey was inadequate to develop a robust sample there, but hesitated to shift our survey towards other towns with higher percentages of English speaking households because of the importance of understanding issues facing the Spanish-speaking community. Instead, we obtained some modest support from the University of Colorado Denver to hire an experienced translator to develop a Spanish version of our household survey instrument, consent forms, and associated recruitment materials. We have also hired two Spanish-speaking graduate research assistants to assist with survey enumeration, translation, and Spanish-language recruitment. After several return visits to Milliken, our Spanish-language graduate assistants were able to survey 6 households, representing a significant portion of the Spanish-only households affected by the floods in Milliken (approximately 15). [2]¹²

Another complication with our survey effort was that one key aid organizations who had offered their support in contacting affected households discovered that their funding requirements prevented them from assisting us. As a result, our progress on recruiting households has been slower than expected. Contacting homeowners has been relatively easy and straightforward, given their fixed address and ongoing communication with town officials about resources like the Hazard Mitigation Grant Program’s “buyout” program. Mobile home park residents, however, most of whose trailers were destroyed and who are thus displaced outside of Lyons and Milliken, were more difficult to locate and survey. The process has been slow but steady; each interview leads to another, and we have slowly accumulated a representative sample from both Lyons and Milliken. Although our sample is small for quantitative purposes, the design of our instrument allowed us to gather detailed explanations directly related to our research questions on access to resources, which resources, challenges, social connections, civic participation, etc.

After completing each survey we entered them into a Microsoft Access database and then exported the data into Excel for analysis.

Results

The following section will describe the preliminary findings from our study, in two parts. The first will discuss our findings from the informational interviews we conducted with local, state, and public officials, as well as some observations from our site visits and participation in flood-related meetings and events. These preliminary observations both informed the content of our survey instrument and will guide our ongoing interviews. The second section will summarize the findings of our baseline survey of flood-impacted households.

Mismatch between grant programs and needs: In meetings with public officials in both Lyons and Milliken, as well as in conversations with state and federal officials in charge of aid programs, there was a consistent feeling that the timelines for various recovery resources were mismatched with the needs of the towns and their residents. This was particularly true in regards to recovery resources, where dozens of grants and sources of funding compete with varying requirements, timelines, reporting requirements, etc. Portions of the largest grant programs, the Hazard Mitigation Grant Program and the Community Development Block Grant Recovery Assistance Program, have yet to be dispensed nine months post-flood. In each case, both the length of time to deliver aid and the State’s decision to change requirements and areas of focus has created a great deal of uncertainty for households and public officials at the local level. Because of the limited funding and guidelines for prioritization of funding, town and county staff members have had to spend hours negotiating with one another to work-out how they would mesh their grant applications so each jurisdiction can receive as much as possible, and at the right time, given the limited funds. A common way of describing the situation on the part of residents and public officials is “limbo;” with so many resources still hanging in the balance, numerous critical decisions are yet to be made.

The mismatch particularly harms lower income households. Some have reported ineligibility for social supports, such as SNAP or Free Lunches because they are living in temporary housing and cannot provide an address or evidence of need and rent payments. The immediate grant programs are not designed to rapidly replace housing with affordable rents. Thus, households who had affordable housing prior to the flood cannot find similarly affordable housing. Many lived in mobile homes and finding a location to place to a new mobile home—if they received a grant to do so—is hampered by town and county restrictions on mobile homes, as well as newly drawn or uncertain flood plain boundaries. A second source of affordable housing was detached housing by independent landlords with single or small inventories. Because these landlords were unlikely to have flood insurance on these units—though they may have had them on their primary homes, they have raised the rents to cover their repair costs. Participants reported rents of $1,200 or more for single bedrooms in Lyons. Grants to replace affordable housing beyond FEMA rental subsidies and mobile home repair and replacement will not become available for several months and housing construction will take another two to three years. Further, FEMA is no longer providing subsidies for temporary trailers so many lower income households have been and expect to be in temporary housing situations for an uncertain amount of time.

Small towns struggle with recovery: In many different ways, small towns and cities struggle with disaster recovery, and this has the potential to put them at a competitive disadvantage for the very resources they need (and hope) to deliver to their residents. Some of the reasons are obvious; municipalities with just a handful of regular staff, including few if any in planning, are poorly equipped to manage the sudden influx of money, volunteers, technical assistance, and media attention that accompanies a major disaster. Similarly, the work required for recovery, especially in communities with heavy infrastructural or housing losses, quickly overwhelms small staffs that are largely for town administration, regulations, and infrastructure maintenance. In normal circumstances, small towns do not employ people with planning, development, and major reconstruction experience; these skills are usually contracted on an as-needed basis. While the state and federal government have expended an enormous amount of money on the recovery, additional support to actually hire more staff has been slow in coming. Towns have received in-kind grants from the State to hire some temporary staff, but are still waiting or just receiving funding to hire long-term staff and professionals to manage the recovery.

Role of local government in recovery varies widely across towns and counties: Local governments view their role in recovery very differently, with potentially important consequences for households. Some communities take a proactive stance towards resident interests and consider themselves the natural advocate of, and conduit for recovery resources of all types. Other local governments see their role as limited to issues like infrastructure, land-use regulation, and management of public resources, and see a larger leadership role for non-profits, the private sector and individual initiative in recovery. Towns also vary on their level of civic engagement before the flood and the visions stated in their comprehensive plans. If a town identified itself as a close community that valued economic diversity before the flood, then these values can be more readily adopted into the recovery process. How these differing views on government affect the long-term recovery of communities will be a key question in our long-term study.

The emotional toll of recovery affects households as well as local leadership: Not surprisingly, disaster recovery takes a heavy emotional toll on both affected households and local leadership (who are sometimes one and the same). After an initial burst of energy and enthusiasm, the long “grind” of recovery starts to take over. By the six-month mark, town staff were increasingly overwhelmed by the requirements of the work, with dozens of meetings each week and a constant stream of grant deadlines, outside organizations and agencies offering assistance, public meetings, and distressed residents, many in difficult and uncertain situations the town is unable to solve, at least immediately. Households are facing a similar kind of fatigue, and even nine months after the flood, many report uncertainty about when, if ever, and how, they will be able to fully recover. Despite applications to recovery programs, e.g. 404, FEMA, and others, and hours of time spent on recovery each day and week, after nine months households with fewer means can still be in limbo with the potential for recovery hinging on several unpredictable factors.

Regional trends often shape local recovery conditions: Finally, our interviews and surveys have made clear the importance of regional trends in shaping local recovery conditions. In Colorado and particularly in places like Lyons and Milliken, rapid population growth has pushed housing prices to a historic high. In Boulder County, where the town of Lyons is located, rental vacancies are at a 12-year low and the median sale price of a single-family home is now over $375,000 (Lindenstein 2014)¹³. In Weld County, where Milliken is located, rental vacancies are some of the lowest in the state due to the strengthening regional economy and growth of the oil and gas industry (Svaldi 2013)¹⁴. Residents who are displaced from affordable housing are at risk of being permanently displaced from their communities, unless local, county or state authorities seek to replace or expand affordable housing opportunities post-flood. Yet, funds to build affordable housing were already over-subscribed before the flood. A key objective in our longer study will be to understand what factors shape decision-making about affordable housing at a local and county level. Another regional trend with significant implications for recovery are long-range controls on urban growth in places like Boulder County, where county-controlled open space effectively jacket small towns and restrict the supply of developable land suitable for building replacement housing.

Baseline Survey Results

Following are some of the initial findings of our household survey (N=27) intended to establish a baseline understanding of flood recovery. We plan to complete a detailed analysis of survey results in September 2014, after we develop a more robust sample that will allow us to do cross-community comparative analysis. Here we report some basic findings about the socio-economic makeup of affected households; the impacts of the flood on employment, housing, and other daily resources; the sources of information households rely on; and some of the most challenging aspects of recovery as described by survey respondents.

Survey participants: Our baseline survey included 27 households in Lyons and Milliken, representing a total of 69 people. The median household size was three. Ten households were headed by a legally married couple living together; one respondent was legally separated, four lived with their partner, five were divorced, three were widowed, and three were single and never married. Of the 69 persons represented in the survey, there were 20 children (under the age of 18), 39 adults, and 8 senior citizens (65 years or older). Thirty-seven reported being Hispanic, Latino or Spanish in origin, 28 were non-Hispanic whites, and 2 were Native American or Alaskan Indian.

Income: The median household income range for all households was $30,000-$39,999, with four households earning less than $10,000 per year and only one household earning more than $100,000 per year (see Table 2). A large majority of our sample is low-income relative to area incomes; according to Colorado State Demography Office, the median household income for Lyons from the 2008-2012 American Community Survey was $74,375 and the median household income for Milliken was $68,288. [3]¹⁵

Housing type & damage: Eight households lived in single-family homes prior to the flood and 17 lived in mobile homes. One household lived in an accessory dwelling unit. Eleven respondents (40.7%) reported that the house/home they were living in at the time of the flood is damaged but repairable while 13 homes were destroyed (48%). The remaining respondents were unsure as to the extent of the damage.

Housing tenure: The vast majority of households in our survey had lived in their community for more than 10 years prior to the flood. Overall, the average tenure in Lyons or Milliken was 15 years pre-flood, with a median tenure of 13.5 years. In Lyons, the mean tenure was 14.5 years, with a median of 10 years. In Milliken, the average tenure was 15.8 years, while the median was 15 years. In both cases, these numbers suggest that a common perception of mobile home park residents, that they are more transient than their neighbors in homes or multi-family housing, is not accurate in these two cases. We acknowledge that our sample might be skewed towards long-term residents, however, given that the vast majority of respondents were still living in the town or nearby area, whereas more recent arrivals might have left the area after the flood because they lacked roots in the community. We should know more once our overall sample grows to include a broader cross-section of households.

Disruption in employment: The majority of households experienced some type of disruption in a member’s employment as a result of the flood. While nearly 40% experienced no disruption, 38.5% of households had a member get reduced hours at work, and 34.6% experienced a period of unemployment (see table 3).

Impact on expenses: Twenty households (74.1%) reported that their expenses (such as mortgage, rent, transportation, or day care) had increased since the flood. The most common expenses to increase were housing costs and transportation costs. Increased housing costs were particularly common among mobile home owners, who typically paid less than $500 per month for ground rent, significantly less than the cost of renting or owning an equivalent square footage on the open market in Lyons or elsewhere in Boulder County. Seven households (27%) reported missing a payment on their mortgage, rent, business, or other debt as a direct result of the flood.

Time spent on flood related work: Even 6-9 months after the flood, some households continue to spend a significant amount of time each week dealing with flood related paperwork, questions, phone calls, meetings, etc. There were significant differences in the experience of households, however; while 2 households reported spending more than 30 hours per week on flood related work, 14 said that they spend 5 hours or less per week. The average number of hours per week across all households who responded (N=26) was 8.3 hours, with a median of 5 hours. In numerous surveys the respondents made a point of saying that immediately after the flood, dealing with FEMA, paperwork, grant applications, insurance companies, and the dozens of other flood-related tasks was a “full-time job.”

Sources of information: households rely on many different sources for information about flood recovery, the most common of which is word of mouth from family, friends, and neighbors (table 4). For Spanish-speaking families, the sources of information were more limited, since Lyons and Milliken do not have local Spanish media. The distribution of information sources shows that multiple channels of communication are important to reach everyone. Because word of mouth was the most common, it indicates some are not accessing other channels. It also implies that local officials, nonprofits, and others should use methods to make sure the word of mouth information is accurate and wide-spread. For instance, those who attend meetings, should be given ideas or handouts on how to share information with those who don’t. Agencies could also identify where most of the word of mouth information is being shared, e.g. coffee shops, churches, neighborhood water sites, etc.

Status of recovery: When asked how long respondents expected it would be until they were fully recovered, only one household reported that they already felt recovered. Fourteen (53.8%) said that they “did not know,” while another three thought it would take at least two years. Several households reported that while their physical recovery was well underway, the mental and emotional effects of the flood were still present. Several parents reported that their children were still having nightmares about the flood, and three different respondents described how rainstorms trigger feelings of fear and anxiety in themselves, their family members, or their neighbors.

Most difficult aspect of recovery: We also asked respondents to discuss the “most challenging part of the recovery process.” Here are a few of their responses:

• The unknown – “not knowing if you’ll ever be able to go back to your home;”
• The uncertainty…no future. If owner of mobile home park sells it, there will be nowhere for us to go. Lyons used to be affordable, now it is half-million dollar homes and yoga studios.
• Getting home, without a doubt. So much uncertainty…
• Getting back into house.
• Finding a replacement home, a permanent solution…something to call home.
• The flow of information…getting accurate information.
• Children are having challenges – their best friends moved away, their parent(s) are without a home. Youngest is very angry and having trouble at school.

Discussion and Conclusion

While we plan to build significantly from the foundation laid with support from the Quick Response Grant, several of our preliminary findings have relevance for disaster recovery policies and planning. At both the community and household level, one of the most common themes of recovery is a sense of being overwhelmed; by the flood of paperwork, volunteer organizations, phone calls, meetings, grant deadlines, policy and planning decisions, and the thousands of other details facing communities like Lyons and Milliken and the households living there. Being overwhelmed takes a very real toll on flood victims and key decision makers, and makes the argument for a recovery resource system that is more efficient, less redundant, and less bureaucratic.

We also find that affordable housing is central in shaping long-term recovery. As the Colorado floods demonstrate, many affordable housing units are located in high hazard areas, like mobile home parks or public housing projects. When those units are damaged or destroyed, especially in a region with rapidly rising housing costs, households are left with few options to return to. As described earlier, these are households with long histories and deep roots in their communities, and sudden severing of ties has long-term implications for flood victims and the towns they live in. Rental assistance has helped numerous households to remain in the county or region, but longer-term solutions are elusive. In order to prevent a local and regional disaster gentrification, policy-makers need to focus on delivering more flexible and effective housing assistance and resources.

References