Relationships Between Distribution of Disaster Aid, Poverty, and Health in Puerto Rico

Publication Date: 2021

Executive Summary

Introduction

Recent research demonstrates U.S. federal disaster aid accelerates economic inequality and increases poverty in the 50 states. Our study explored whether this was also true in Puerto Rico. Although Puerto Rico is a part of the United States and receives federal aid, its status as an unincorporated territory makes the process distinct. We investigated how demographic, cultural, and bureaucratic distinctions influenced residents’ lived realities after Hurricanes Irma and María struck Puerto Rico in 2017. Our guiding question was: How did the disbursement of disaster aid impact relationships between hurricane damages, poverty, and population vulnerability to COVID-19?

Research Design

We used a mixed method approach that involved examining administrative data and interviewing residents. First, we investigated how poverty rates, hurricane damages, and disaster aid were correlated. Next, we conducted case studies in two municipalities that we selected based on the findings of our quantitative research. In both municipalities, we interviewed a total of four public employees, four business owners, and 71 residents (where N=76, because three business owners were also residents). Then we examined correlations between property damages, hurricane fatalities, and the way economic inequality and COVID-19 infection rates were affected by the way disaster aid was disbursed.

Preliminary Findings

After accounting for changes in population, we found that municipal poverty rates began to increase at a faster rate after the hurricanes. We found this increase positively correlated with hurricane fatalities but not property damage. Moreover, poverty accelerated faster in areas that received more disaster aid. Case studies provided a disaggregated view of disaster aid revealing its unequal distribution. Aid remained out of reach for already vulnerable populations. The unequal distribution of aid increased health inequities. Interviews with residents highlighted the growth of deep poverty. In case studies, we found similarities between the two municipalities, such as an overall sense of violence from bureaucracy and governmental neglect, defined in our analysis as structural violence, thus emphasizing the health costs of delayed and inequitable disbursement of government money. We also found differences, such as weaker social support networks where poverty was less extreme. This resulted in unmet needs for assistance and reiterated experiences of neglect. In contrast, where deep poverty was an existing and growing problem, hyperlocal support networks created after the hurricanes remained in place and were more easily activated in response to the COVID-19 pandemic. Lastly, we found the cumulative number of COVID-19 cases to be positively correlated with each of the following ordered from strongest correlation to weakest: disbursed disaster aid, hurricane fatalities, economic inequality, and hurricane property damages.

Implications

There are several policy implications of our findings. Top among them is to immediately adopt an equity framework in all disaster responses and policy making for disaster recovery. We found a need to develop an equity-conscious approach to aid distribution so that more people can receive it. The top priority should be to make aid easier to access for populations facing compound economic, social, and health challenges. Our interviews revealed a recursive relationship between public health and disaster recovery that is not addressed in policy. Public health experts, social workers, and community representatives should collaborate on policy design. Decisions about disaster aid should take into consideration health metrics in distribution strategies. We also recommend examining assumptions underlying the use of Small Business Association (SBA) loans as an alternative to aid for rural households. Aid programs should be regularly monitored and evaluated to ensure that public assistance is distributed equitably to the diverse racial and ethnic groups living in Puerto Rico, and that it is also distributed fairly across geographic regions. Hazard assessments can be used to prevent growth in poverty by accounting for the relative cost of homes and livelihoods for rural populations. The ownership documentation requirement in the hazard damage assessment disqualified many of the people most in need of aid. These requirements should be reviewed.

Conclusions

Our findings reinforce and build upon those of Howell and Elliott (2018¹) while also revealing the need for additional research. We recommend investigating how post-disaster changes in organizational ecosystems affected economic and health outcomes. After Hurricane María, the number of non-profit organizations in Puerto Rico increased by roughly threefold. Researchers in the United States have found that growth in social organizations increased poverty, with the exception of advocacy organizations (Smiley et al., 2018²). There is a need to better understand the impacts of differing post-disaster recovery inputs—such as public and private financial aid, material goods, or loans. Future research also should address more thoroughly complex factors such as the repercussions of long-term austerity measures, tax structures created to attract immigration of wealthy individuals, and incentivized post-disaster out-migration of the economically disadvantaged. We hypothesize that each of these patterns impacts health outcomes and that disasters in turn change these social and economic relationships.

Compound natural disasters, such as Hurricanes Irma and María, followed by the COVID-19 pandemic, are increasing in frequency and intensity across the world. It is imperative that we understand the interactions between these not-so-natural disasters and our human systems—from neighborhood organizations to global economies. We also need to pay attention to how these interactions affect growing economic and health inequities.

Introduction and Literature Review

Geographic Context

Puerto Rico is an unincorporated territory of the United States located in the Caribbean Sea. It consists of three inhabited islands. The largest island is divided into 76 municipalities, and the two smaller islands are each their own municipalities bringing the total number to 78. The poverty rate in Puerto Rico is by far the highest in the nation. The number of children in Puerto Rico growing up in high poverty areas is six times that of the poorest jurisdiction in the 50 states (Backiel, 2015³). In addition to poverty, Puerto Rico has very high economic inequality. Its Gini-coefficient is estimated to be 0.55 (Colón Reyes, 2019), which, if compared with sovereign nations of the world, would put it among the top ten most unequal (World Bank, 2020⁴).

Disaster Context

In the past four years, Puerto Rico has experienced two Category 5 hurricanes (Irma and María in September 2017), thousands of earthquakes reaching up to magnitude 6.4 on the Richter scale (December 2019 to early 2020), and a pandemic (COVID-19, with first reported cases in early March 2020). A compound disaster is, “an emergency situation with adverse consequences resulting from different but related disaster-agents” (Wachira, 1997⁵). Thus, in Puerto Rico, the two hurricanes of 2017 could be considered a compound disaster as could the simultaneous hazards of the earthquake sequence and the COVID-19 pandemic in early 2020. Though historically high levels of federal disaster aid were approved for Puerto Rico after the hurricanes, historically low proportions of aid have been disbursed to this day almost four years after the disasters. As of March 2021, only 27% of the over $67 billion allocated had been disbursed (COR-3, 2021⁶), and only 26% of Federal Emergency Management Agency (FEMA) funds sent to Puerto Rico had been disbursed to municipalities (Ruiz Kuilan, 2021⁷). The delayed recovery aid meant that the hazards of 2020 added to the vulnerabilities of a population that was still experiencing health and property damages from the dual hurricanes of 2017, creating a cascading disaster. Pescaroli and Alexander (2016) explain, “cascading disasters tend to highlight unresolved vulnerabilities in human society.” In the case of Puerto Rico, the meteorological, geophysical, and biological disasters followed the political and economic disaster of bankruptcy in 2015, which was used as justification for austerity measures that have resulted in increased fatalities, growing health inequities, and threatened efforts to prepare for each subsequent disaster.

Early evidence of the health costs of cascading disasters in Puerto Rico can be found in the accounting of deaths before and after Hurricane María. Benach et al. (2019⁸) found that in 2016, after the bankruptcy declaration, excess deaths were registered. Furthermore, in the aftermath of Hurricane María, “people living in poor municipalities were 60% more at risk of dying months later due to the hurricane.” The economic crisis also increased vulnerability to disease in the population, where the lack of public health investment and the economically motivated emigration—migration out from Puerto Rico to other parts of the United States and the world—of thousands of health professionals has increased incidence of chronic diseases (Parés, 2016⁹), as well as risks for COVID-19 complications and death. When COVID-19 was first detected in Puerto Rico, it had arrived in three small Caribbean islands where 20% of the population was elderly and specific health vulnerabilities were high including a 16% diabetes rate (Garriga-López, 2020¹⁰). Puerto Rico’s population is majority Latine^{Endnote 1}, and while both college education and homeownership are higher in parts of Puerto Rico than in some U.S. counties, neither are as protective against poverty in Puerto Rico where close to half of the population lives in poverty.

Literature Review

Poverty and social inequity are longstanding and growing problems that are currently at their highest points in human history (Sen, 1983¹¹; Kohler, 2017). Research on causes of this trend point to multiple feedback loops. Health is one such loop: poverty worsens population health, and decreased population health worsens poverty. Then poverty continues to erode population health even further. Similarly, disaster damages increase poverty, which further increases vulnerability to future natural disasters. Our study examined the idea that aid programs as currently designed and implemented exacerbate cycles of increasing poverty and diminishing health rather than helping to solve those problems.

The ecosocial theoretical framework created by Krieger (2001¹²) is useful for exploring and visualizing how public policies, culture, organizations, and other social determinants work together to impact health at the community and individual levels. There are many detrimental social determinants of health at play in Puerto Rico, from a failing health system (Portela & Sommers, 2015¹³), to inequalities in access to data (Rodríguez-Vilá et al., 2017¹⁴), to widespread exclusion from formal markets, social disenfranchisement, population displacement, changing access to natural resources due to coastal erosion, and limited access to water and energy resources. Recent post-disaster research in Puerto Rico points to the relationship between health impacts, social inequality, and natural disasters. Flores and Arroyo Quijano (2017) report both an increase in post-traumatic stress disorder (health outcome) and a decrease in academic performance (social determinant of health) in the children as a result of 2017 hurricanes. On the other end of the age spectrum, García et al. (2020¹⁵) found that for older adults the compound disasters of the 2017 hurricanes and the bankruptcy contributed both to worse health outcomes due to poor healthcare, and to increased risks of adverse COVID-19 outcomes, due to the impacts on several social determinants of health including increasing wealth inequality and weakening social support networks due to emigration.

Public health researchers identify social inequities as a fundamental cause of inequities in health outcomes, or a “cause of causes” (Phelan & Link, 2013¹⁶). Phelan and Link showed how poverty reduces access to the resources, including technologies and capitals, that people can use to protect or improve their health. We extend this analysis to disaster readiness, response, and recovery. Impoverished people and communities have less access to the resources they need to mitigate hazard impacts, and disasters intensify this inequity leading to damaged health, property, and community functioning that further decreases access to resources and makes preparing for the next disaster more difficult. Natural hazard damages impact individual and community health both directly and indirectly, so it is essential to consider the social determinants of health in addition to the direct health impacts of a disaster to understand how they increase vulnerability to future risks.

COVID-19, like many threats to health, thrives where poverty and income inequality are high. Patel et al. (2020¹⁷) have shown that lower socioeconomic status could increase one’s risk for COVID-19 infection via numerous mechanisms. The social disadvantages created by systemic racism in the United States have led to increased rates of infection and mortality that have devastated communities of color. Rodríguez-Díaz et al. (2020¹⁸) found that majority Latine counties across the United States experienced a higher COVID-19 infection rate at 90.9 cases versus 82 cases per 100,000 population (they included Puerto Rico as one county). Blundell et al. (2020¹⁹) describe patterns pointing to the likelihood that the pandemic will become a vector for increased social inequities. As of July 9, 2021, Puerto Rico has experienced 2,553 deaths (0.8 per 100,000 mortality rate) and more than 174,000 positive cases (The New York Times, 2021²⁰) (74.8 per 100,000 infection rate). Research is needed to examine how the celerity of Hurricane María recovery funds are disbursed and how other social determinants may have impacted COVID-19 infection rates in Puerto Rico.

Howell and Elliott (2018) found that federal disaster aid was associated with increased economic inequality across all counties of the United States. They asserted that aid increased poverty and wealth disparities in part because its focus on property damages instead of individual and social wellness. The concept of social capital may be helpful in explaining the mechanisms that underlie this process. Talbot et al. (2020²¹) studied bridging and bonding social capital's ^{Endnote 2} impact on informal reconstruction activity in two municipalities of Puerto Rico after Hurricane María. Their findings led them to consider this possibility: “bridging social capital may provide access to resources that help with initial reconstruction but do not enhance long-term recovery.” It may be that bridging social capital is helpful during the response period but not in the recovery period when linking social capital may become paramount. It is especially important to understand which strategies are most effective in the context of cascading disasters because they collapse the boundaries between response, recovery, and readiness.

To further understand these differences, we drew upon the work of Hernández et al. (2018²²). In their exploration of the impact of Hurricane Sandy on residents of public housing, they developed the conceptual framework of the “resilience reserve.” The concept helps to explain why people in low-resource communities might cooperate for emergency response activity in the early stages of a disaster but as the reserve gets depleted, and outside resources do not come to them, the energy for such activity wanes. Sociologist Linda Colón emphasized that poverty is not only measured in economic terms but also in access to political power and other forms of social capital (Colón, forthcoming). An understanding of how social capital influences the access and flow of financial capital during the upheaval caused by natural disasters may reveal potential strategies for more equitable aid distribution.

Study Rationale

We built on scholarship exploring the role of disaster recovery in poverty with the goal of improving economic equality and health equity in communities impacted by natural disasters. We used mixed methods to connect macro and micro frameworks for understanding the impacts of disaster recovery inputs. We contextualized the “resilience reserve” framework within the changes in local economies. Our study can contribute to practice and policy that improves equitable distribution of aid.

Methods

The guiding research question for the study is: How did the disbursement of federal disaster aid after the 2017 hurricanes in Puerto Rico impact the relationships between hazard damages, poverty, and population vulnerability to the public health risk posed by COVID-19? The study design used the convergent framework (Peek et al., 2020²³) for a transdisciplinary approach to address a complex problem, where we investigated the question using four specific aims. The study blends insights from economics, applied anthropology, and public health. We used qualitative and quantitative methods to examine dynamic relationships between hazard damages, emergency responses, recovery efforts, and public health vulnerability in Puerto Rico. For Aims 1-3, correlation and regression models enabled analyses of municipal measures of damages, aid, poverty, economic equality, and COVID-19 burden. For Aim 4, we conducted case studies to explore mechanisms of relationships identified at the macro level. Our qualitative methodology was designed to investigate the socio-economic impacts of the 2017 hurricanes, recovery efforts, and the 2020 COVID-19 pandemic.

Data, Methods, and Procedures

Aim 1

Examine the changing rate of municipal poverty from 2015 to 2019 and whether damages from Hurricanes Irma and María (2017) accelerated increases in poverty.

Poverty was measured as the proportion of the population whose income falls below the U.S. Census poverty line. We estimated the changing rate of municipal poverty by calculating the relationship between year and poverty while holding constant other changing demographic factors including U.S. Census estimates of the total population, proportion of the population with a bachelor's degree, percent of the population below age 18 and above 65, and the Puerto Rico Department of Labor’s quarterly average wage. We then calculated the relationship between year, total population, and poverty for each municipality separately. By calculating the difference between the year coefficient pre-2017 and the year coefficient post-2017, we could approximate how much the change in poverty rate was altered after the hurricanes. Using this as a dependent variable, we examined the relationship between this alteration and hurricane damage, conceptualized as both property damage and fatalities. Hurricane property damage was approximated with the Special Hazards Events and Losses Database for the United States (SHELDUSTM). Non-crop property damages and fatalities were calculated by the Puerto Rico Center for Investigative Reporting.

Aim 2

Ascertain the influence of federal disaster aid on the change in poverty rates.

For Aim 2, we built on the Aim 1 models by adding data from the Federal Emergency Management Agency (FEMA) and Community Development Block Grant Disaster Recovery (CDBG-DR) programs about their aid distribution in each municipality. Our key independent variable for this analysis was total disbursed aid, which included assistance to individual households and assistance to municipalities.

Aim 3

Show the relationships between hurricane damages, disaster aid, economic inequality and each municipality’s ability to prepare for a public health threat by investigating distribution of COVID-19 cases across municipalities.

For Aim 3 we used Puerto Rico’s Department of Health municipal COVID-19 cumulative case counts from April 2020 to April 2021. We included non-duplicated positive Polymerase Chain Reaction and serology tests. We calculated correlation estimates between COVID-19 case counts and total aid disbursed, number of fatalities attributed to Hurricane María, total damages in dollars, and the Gini coefficient for each municipality.

Aim 4

Identify the underlying mechanisms of dynamic relationships identified in Aims 1-3 by exploring the impacts of federal disaster aid in two municipalities using case study methodology.

We used ethnographic observation and structured interviews (N=76) to collect data. We developed interview guides to focus on factors in multiple dimensions of the ecosocial model. We identified areas of interest, curiosity, and confusion for further exploration in the process of discussing results of Aims 1 and 2.

Study Sites

The research location for all aims was Puerto Rico. For Aims 1–3, we compared data for all 78 municipalities. Results from Aim 2 guided our case study site selection for Aim 4. We identified the range of correlations between disbursed aid and changes in poverty in all municipalities. Then we selected one of the three municipalities that was closest to the average correlation and the municipality with the farthest outlier correlation (which happened to be the smallest). All correlations were positive and clustered around the averages, leading us to believe that conducting a case study in one of the three towns with the average correlation could reveal broader underlying mechanisms that contribute to the positive relationship between federal aid distribution and increasing poverty. In addition, we felt a comparison with the town that had the greatest outlier correlation would help us to identify mediating factors that might be reducing the intensity of the relationship within that town. We hoped that our observations would lead to recommendations for interventions that could be used on a broad scale. We also wanted to deepen our understanding of the interaction between all factors in the town where federal aid seemed to have the smallest impact on increasing poverty. Since there were three municipalities with the average correlation, we were able to select two municipalities in the same peri-urban region.

Primary data collection for Aim 4 was conducted in person in the two selected municipalities. In order to maintain privacy and honor our confidentiality agreements we refer to the sites with the fictitious names Nube and Suelo. Nube is a town with a population of under 40,000 people and is described by residents as campo (rural). Nube represents the average positive relationship between aid and poverty in Puerto Rico. In Nube, over a seven-year period, the percentage of the population living below poverty (PPBPL) grew by 4%. Suelo, on the other hand, is the town with the smallest identified relationship between aid and change in poverty (although it is still a positive relationship). Its population is about 70,000 and has both rural communities and suburban developments. In Suelo the PPBPL decreased by 19% over the same period.

Sample Size and Participants

For Aims 1-3, data from all municipalities were used. As discussed above, we selected two theoretically advantageous municipalities for the Aim 4 case studies. Within each municipality, we sought study participants in two categories: 1) unaffiliated residents and 2) people affiliated with selected public and private organizations (both nonprofit and for-profit) either as staff, volunteers, or service customers/clients. For the former category, invitations to participate in the research were distributed on social media and handed out to individuals in public spaces. Eligibility criteria for unaffiliated residents to participate included age 18 years or older and residence in the town for at least five years. For the latter category, we conducted stratified random sampling from the sample frame of Puerto Rico State Department records, to select 30 organizations. We then invited organizational leaders from selected organizations to participate by connecting us to affiliated individuals who then received individual invitations to participate.

Efforts to secure local residents as research assistants proved effective for securing participation ^{Endnote 3}. Our final sample included 20 organization-affiliated participants (four business owners, four public servants, and 12 employees or social organization members), and 56 unaffiliated residents. The poverty rates of those interviewed reflected the overall poverty in each municipality. In Nube the percentage of the population earning incomes below poverty level was 50-59% and in our sample it was 58%; in Suelo the municipal poverty rate was 26-39% and 32% of our sample participants reported earning less than $20,000 annually. For further information on participant demographics by municipality, see Table 1.

Data Analysis

For Aims 1, 2, and 3 we estimated panel and cross-sectional regression models and correlations. See Appendix A for the equations. For Aim 4, we used a hybrid inductive/deductive thematic analysis technique outlined by Fereday and Muir-Cochrane (2006²⁴) to iteratively develop and test theory. All data collectors identified recurring or prevalent themes in the interviews they conducted. We transcribed 39% of interviews and conducted language analysis. We triangulated findings from the computer-assisted language analysis with ethnographic observations and direct text analysis (Wignall & Barry, 2019²⁵) exploring tensions and contradictions, needs, and agency. Identified themes were defined and organized along eco-social dimensions (see Figures 1 and 2). Four members of the team collaboratively engaged in thematic analysis. Next, we compared our qualitative findings between communities and to our findings from Aims 1-3 to look for fit and contradictions.

Researcher Positionality, Reciprocity, and Other Ethical Considerations

Our transdisciplinary research team, which consisted of an economist, applied anthropologist, and public health researcher-practitioner, collectively developed and guided the study with the support of 10 Puerto Rican research assistants and two senior consultants ^{Endnote 4}. All co-authors live in Puerto Rico and are bilingual in Spanish and English, two are Puerto Rican. Our team was mindful of the history of abusive research in Puerto Rico and we are committed not only to ensuring ethical treatment of participants but also to establishing reciprocity with them. This meant communicating informed consent in understandable language, ensuring privacy, reducing risk of COVID-19 transmission during research activities, protecting and anonymizing data, and compensating participants for their time with cash. We also will ensure that all participants are able to access and understand our findings, by hosting two in-person community events, one each in Nube and Suelo, and disseminating a short pamphlet communicating findings in sixth grade language and with visual graphics. In our study design and dissemination, we actively integrated awareness of the potential impacts of our study on the current narrative discourse around disaster funding. All primary data collection procedures were reviewed and approved by an external entity, Ethical & Independent Review Services, and all participants provided informed consent before participation. Disaster-zone research does not yet have a unifying code of conduct (Peek et al., 2020), all researchers participated in an online training series to ensure the team understood the ethical ramifications of our research. The series was titled, “Converge Training on Disaster Research,” and it was organized by the CONVERGE facility at the University of Colorado Natural Hazards Center and supported by the Centers for Disease Control and Prevention and the National Science Foundation.

Lessons Learned

Secondary data gathering met with social, political, and economic challenges that highlight the inconsistencies in data management across the United States. When we consulted federal data sources for Puerto Rico we confronted the reality that data from U.S. territories is treated differently across agencies and even within agencies. In some reports municipalities are treated as counties while in others counties are treated as regions of municipalities. Some reports are not offered at all, creating “data deserts.” Our experience suggests this may be the case for all U.S. territories, which are often left out of databases that tally states but not territories. Personal appeals to government agencies for better data ran into the obstacle of post-election administration changes at the federal, territory, and municipal levels. For the data we did get, we found it important to “trust but verify.” For example, we obtained the urban-rural measure from the National Center for Health Statistics that other researchers rely on to describe the degree of urbanity/rurality in a community as this geographic distinction greatly impacts many social, economic, and health outcomes. Upon inspection, however, we realized that the way the data categories were applied to Puerto Rico did not reflect a realistic understanding of local geography. To solve this problem, we created a population density measure that is imperfect but, we feel, better captures the true impact on infectious disease risk.

Findings

Aim 1

As noted by other researchers, poverty rates in Puerto Rico are considerably higher than the rates in the 50 United States—with nearly half of the residents living under the federal poverty rate. The poverty rate fluctuates from year to year (see Table 2). Likewise, municipal average wages also have fluctuated. The total population of Puerto Rico is also steadily declining. The population over 65 is increasing at the same time the population under 18 years of age is decreasing. These statistics help us understand how poverty in Puerto Rico is changing over time. To understand whether the change in poverty rates was altered by the 2017 hurricanes, we estimated fixed effects panel models examining the change in poverty pre- and post-hurricane. Holding constant the other demographic changes, poverty decreased from 2015 to 2017 (as demonstrated by the year coefficient in Table 3). Yet, after the hurricanes, this relationship flipped ^{Endnote 5}. Holding constant the changes in population, poverty began increasing every year. This national average poverty rate mirrors previous research demonstrating disasters increase poverty. However, this summary measure is unable to take into consideration the specific level of hurricane damage in the municipality.

To understand the relationship between hurricane damage and increases in poverty, we created a fixed-effects model for each municipality pre- and post-hurricane. We used the municipal specific coefficients to estimate a local change in poverty rates (see Table 4 for descriptive parameters). Using this difference as our dependent variable, we explored how hurricane damage has changed the poverty rate over time, holding constant the municipalities’ demographics when the hurricanes hit. We used two estimates of hurricane damages—property damages per capita and fatalities per capita.

As seen in Table 5, property damage caused by the hurricanes did not help us explain the changing rate in poverty shown in Table 3. In fact, property damage is inversely related to increases in poverty. However, fatalities per capita have a strong positive relationship with the change in poverty rate over time. This suggests municipalities that suffered the highest human toll from the hurricanes experienced a long-term disaster that caused poverty to steeply rise. Additionally, looking at the weak relationship between property damages and fatalities, it is evident that higher property damages are more reflective of higher property values in richer municipalities than the level of physical destruction.

Aim 2

Building on Aim 1 findings, we examined how disaster aid alters the poverty rate post-hurricanes. Holding constant the fatalities, we found disaster aid was positively associated with increasing poverty rates. In other words, more aid accelerated increases in poverty.

Aim 3

We found positive correlations between each of the variables in Aim 3. The highest correlation with COVID-19 cases was with the total amount of aid disbursed. This was followed by a high correlation between COVID-19 cases and the number of fatalities from Hurricane María. COVID-19 cases were also positively correlated with the Gini coefficients for each municipality. Finally, the correlation between COVID-19 cases and the total amount of property damages in millions of dollars was the lowest, although it was also still positive. See Tables 6 and 7 for descriptive statistics and the correlation matrix.

Aim 4

Tabulated answers to disaster experience questions are presented in Table 8. While the portion of the sample that experienced total loss (10%) was lower than the estimate for all of Puerto Rico (16%), we saw a marked difference between towns. Thirteen percent of Nube participants experienced total loss but only 4% of Suelo participants reported total loss. This imbalance was reflected in other measures as well: we tallied all participant reports of devastated communities, and found that 71% came from participants located in Nube; similarly, 70% of all participant reports of loss of life or severe impact to health of a family member were reported by Nube participants. As expected, participants from Suelo, which suffered less damage, received less disaster aid. The type and amount of aid received by participants from Suelo does not conform to expectations. Higher financial aid was reported by Suelo participants, and non-monetary aid was easier to access in Nube. To put this into context, government reports show the total disbursed aid to date was $25.2 million (70% direct household assistance) to Nube. In Suelo, total disbursed aid was $34.2 million (83% to households) (COR-3, 2021).

Participant assessments of the disaster response and recovery across four eco-social dimensions are represented in two models—one for each town, Figures 1 and 2. The models aim to visualize the difference between sources of disaster recovery activities, actors and emotional support that participants mentioned as being important to their own recovery, with dimensions that participants in each town relied on more being represented with thicker layers. In Nube, the most sustaining sources of aid were at the levels of the local community and interpersonal relationships. Many such actors remained active even until and beyond the date of data collection, in March 2021. In contrast in Suelo, disaster recovery activities at the community and relationship levels were less robust. More disaster response and recovery activities in Suelo started at the institutional and individual levels.

Further detail on Aim 4 findings is provided in Appendix B for the codebook and Appendix C for theme definitions. These appendices include computer-assisted textual analysis as well as definitions with quotes from our interviews for each theme. A theme that arose from textual analysis and was not in our a priori framework was the “violence of bureaucracy,” which we define as violence that is ascribed to the government against the people in its care. Additionally, textual analysis reveals repeated connections between health, well-being, and the hardships endured as a result of hazard damages. We identified other salient themes including the strengthening of local connections, the manifestation of the resilience reserve among neighbors, and the social capital that accumulates between nongovernmental actors. An example of social capital and how it impacted economic well-being was described in an interview with a widow who lived alone and was able to sell alcapurrias (Puerto Rican fried fritters) she made using green bananas that her neighbor with a farm gave her regularly. His surplus was her opportunity to supplement her income and begin the repairs to her home. Another example arose in an interview with a woman who was both a seamstress and mother, who lost her materials and tools (fabrics, sewing machines, patterns) to the hurricane. When COVID-19 arrived in Puerto Rico, the woman was so concerned about the health of her son, who was an essential worker, that she sewed him a dozen masks by hand because none could be found or bought. He started giving them away to friends and colleagues, so she sewed more. Then she began selling them at a low price and eventually made enough money to repair two of her sewing machines.

Our qualitative data showed how poverty in Nube had become more extreme. Social organizations were still active partners, distributing aid, especially food, during the pandemic. Reinforcing findings from Aims 1 and 2, personal stories of participants from both municipalities suggested there is growing economic and social inequality between as well as within municipalities. The stories of neighbors, businesses, and life getting back to normal before the pandemic were easily traced to Suelo. The stories of roads not yet fixed, lack of transportation infrastructure, and the absence of an emergency care facility (until 2019) were typical of Nube.

Across the personal narratives, more value was put on monetary aid. It was clear that monetary aid was needed to enable recovery, and food aid was needed for immediate survival. Sharing food was good, but money to rebuild was better. This assessment is not intended to discount the important emergency response but to advance awareness of the lack of recovery in the face of cascading disasters. Another theme that came across from the interviews was that participants connected the deteriorating mental health of relatives, especially elderly relatives, to quickly declining physical health. This may have made them susceptible to infectious (and other) diseases, such as COVID-19, and even premature death.

Another problem revealed by our qualitative analyses was the unequal distribution of disaster relief funds. It seemed that funds flowed to those who could guarantee loans and prove ownership. An example of this came from a man recruited for an interview at a bar in Suelo, where he was spending a Saturday afternoon with his wife and friend. After months of him trying to get aid for his sister’s home, his son’s home, and his own home, FEMA argued that he was trying to defraud them because the three homes were really one property. In the countryside this happens often. One family builds homes for different members in different corners of one property. He was frustrated. His daughter’s friend who worked for FEMA told him where to go and what to do. He thought he would finally get the needed aid, but months later, indignant, he gave up. Where there are pre-existing high levels of poverty, residents are more susceptible to these kinds of bureaucratic problems that make aid distribution unequal.

Discussion

This study built upon the work of other scholars who, like Howell and Elliott (2018), work from “the twin realization that natural hazards do not just bring damages, they also bring resources; and equal aid is not equitable aid especially when it is systemically designed to restore property rather than communities.” Our Aims 1 and 2 findings expose the growing vulnerability that helps explain the distribution of COVID-19 infections across municipalities, underscoring the urgency of translating these findings into policy changes. Specifically, those parts of Puerto Rico that suffered the highest rates of fatalities during the hurricane and the response period, also suffered the highest rates of increasing poverty during the recovery period. Aim 3 adds to the story, as we learn that fatalities from the hurricanes, as well as federal disaster aid, were correlated with higher COVID-19 case burdens. The correlation with the Gini coefficient showed that economic inequality within each municipality also contributed to COVID-19 population risk.

The De Alwis and Noy (2019²⁶) study of inequality in Sri Lanka following the 2004 tsunami found regional differences in recovery. Higher-income areas with less damage experienced better recovery. This was similar to our findings on the differences in receiving disaster aid experienced by participants in the two case study towns, highlighting the importance of continuing to investigate these relationships at multiple eco-social levels. Suelo is more middle class than Nube, and more participants received FEMA funds that were disbursed to individual households, while fewer homes were damaged and to a lesser degree. Nube, on the other hand, had higher poverty rates prior to 2017, and suffered greater hurricane damage, yet fewer people received FEMA aid.

In addition to the unequal distribution of relief funds, another potential mechanism of the identified relationship between aid and accelerating poverty growth may be the way aid impacted the post-disaster organizational ecosystem. Smiley et al. (2018) found that in the United States not only do hazard damages lead to increased economic inequality one year later, but that the number of private organizations, both non-profit and for-profit, also increases one year after hazard damages. Their research shows that growth in the number of non-profit organizations correlates with increased poverty, with the exception of advocacy organizations. Our case study supports their finding that growth in the number of businesses, on the other hand, is associated with decreased poverty (ibid). The post-disaster social network extends to local businesses and creates economic opportunity. In Nube, more participants mentioned colmados, or minimarkets, as caring businesses, and restaurants that organized community activities. Another researcher noted the impact of hardware stores on the economy and even engineering safety across Puerto Rico in the disaster recovery period (Briar Goldwyn, personal communication, 2021).

We incorporated concepts of social capital into our analysis to better understand how resilience on a household level may connect to organizational and institutional levels. Islam and Walkerden (2015²⁷) found in their investigation of social capital among Bangladeshi households affected by Cyclone Sidr that bridging social capital networks initially protected against loss but that protective effect did not extend into the recovery period. It broke down in the face of competition for external recovery support. They specifically attribute this breakdown to uneven distribution of disaster aid and inaccessible avenues for community participation in the disaster recovery process. Their description of these processes parallels Talbot et al.’s (2020) study of social capital impact on informal post-disaster reconstruction in Puerto Rico. It may provide a clue to the differences we saw in the two towns of our case study. The lack of aid and outside support going to Nube may paradoxically be connected to the enduring strength of hyperlocal support networks even beyond the initial emergency response period.

We also expected to find many residents of Puerto Rico who were impacted by Hurricane María demonstrating a similar depletion of reserves due to the hardships of daily life that were studied by Hernandez et al. (2018). We further hoped to understand whether or how this framework might apply to compound and cascading disasters. In our case studies, we found more veiled and direct references to the resilience reserve behaving as a temporary resource that builds and gets depleted in Suelo. For example, one participant described a strategy where all neighbors contributed to buying materials for everyone’s use but indicated that it only lasted a short time. It could be that the nature of extreme poverty experienced by many participants in Nube created a different sort of resilience reserve than that described by Hernandez and colleagues because incomes were extremely far below the poverty line. It may be that residents of Suelo are more comparable to residents of public housing in New York City, which was the focus population of the Hernandez study. It could be that the chronic hardships endured in Nube require a greater level of connection for everyday survival. Despite greater damages being reported in Nube, residents of Suelo reported a 20% higher incidence of deteriorating mental health compared to residents of Nube. In spite of more FEMA relief aid being directed to Suelo, as well as lower reports of damages, the social capital developed in Nube likely had a protective effect on its residents (in terms of mental health but not necessarily disaster-related fatalities or COVID-19 case rates). This idea was supported by several interviewees who stated that the logistics network used after Hurricane María was quickly renewed in Nube at the start of the COVID-19 pandemic. For example, municipal employees worked with local organizations to deliver a small bag with food and hygiene items (including hand sanitizer) to every house. They worked on average 12-hour days at the beginning of the pandemic.

Key Findings

Comparing poverty trends before and after 2017 in Puerto Rico revealed that decreasing poverty was reversed by the 2017 hurricanes. Poverty trends began increasing the following year. Greater disaster-associated fatalities and larger amounts of disaster aid were both associated with greater acceleration of poverty. Similarly, those two factors were most strongly correlated with more COVID-19 cases across Puerto Rico’s municipalities. In light of Benach et al.’s (2019) description of longstanding channels of compounded inequity in Puerto Rico, our findings suggest that both disaster aid and infectious diseases travel along these same channels and in the process deepen them. Examples include informal construction and compromised construction safety, economic marginalization and involuntary emigration, lack of information access and COVID-19 testing, formal title requirements and FEMA funds, among many others. Our case studies revealed potential mechanisms recreating these relationships. Our qualitative research found similarities across all participants, such as an overall sense of violence from bureaucracy and governmental neglect. We also found differences, such as a fluctuating resilience reserve where poverty was less extreme and more enduring hyperlocal support networks where extreme poverty creates everyday disasters that require constant survival responses. Our findings also support patterns described by Benach et al. (ibid), where the initial threat to survival immediately post-disaster gave way to longer term threats to mental health, which eroded physical health. When the COVID-19 pandemic arrived it compounded and magnified existing inequities.

Policy Implications

The recursive relationship between hazard damages, aid distribution, poverty, and health reinforces the idea that economic policy is emergency preparedness policy is disaster response policy is public health policy. The correlations between COVID-19 case rates and various aspects of poverty are a warning that unless policies are developed and implemented with a conscientious equity strategy they will serve to deepen inequity rather than alleviate it. All policy must be made with the understanding that vulnerability to hazard damages and the ability to recover from disasters are directly shaped by existing socioeconomic and racial inequities. In the case of Puerto Rico, long standing racial inequities are often overlooked and silenced, but they are undeniably present (Rivera-Batiz, 2005²⁸; Rodríguez-Díaz & Lewellen-Williams, 2020²⁹). The pattern of unequal treatment is systemic. Willison et al. (2019³⁰) found that, “within the first nine days after the hurricanes hit, both Harvey and Irma survivors had already each received nearly US$100 million in FEMA dollars awarded to individuals and families, whereas María survivors had only received slightly over US$6 million in recovery aid.” Framed within a national context, the federal government’s treatment of Puerto Rico in its disaster aid disbursement is part of a demonstrated trend in inequitable treatment and outcomes for Latine communities across all other parts of the United States. Our mixed method findings demonstrate that unless equity is conscientiously aimed for, aid is likely to follow existing heavily worn paths toward power and privilege. It will continue to marginalize poor people and amplify existing inequities. Aid disbursement strategies must be purposefully designed to proportionally meet the needs of communities while accounting for pre-existing population vulnerabilities.

Five specific policy recommendations were identified:

1. Develop a differentiated approach to aid distribution to broaden access, monitor diversity, and use data to make regular adjustments. Identified connections between fatalities, economic inequality and infectious disease distribution demonstrate that this is a matter of life or death.
2. Review and revise the FEMA aid application process to make it more accessible to lower-income homeowners, create a budget for application assistance to help the lowest-income homeowners, and expand ownership documentation options. Denials of aid applications were often based on homeowners not having the correct legal documentation. They also lacked the means to fight unjust denials. This approach to aid distribution contributes both to inequity within municipalities and between them, as well as between Puerto Rico and the stateside United States.
3. Adjust hazard damage assessment procedures to ensure that funds provided will enable repairs to be completed in both urban and rural communities. Our participants in rural communities reported being unable to complete basic repairs because they were not awarded enough aid. In our sample, not even one participant who needed a roof repair received the full amount needed to complete the repair so that they could live under an intact, leak-free roof.
4. Review and revise assumptions about using Small Business Administration loans as an alternative to aid for rural households. Our case studies revealed that loans were given in lieu of aid to people who had minimal assets to rely on for survival, even counting the ownership of multiple chickens as a small business. The ease with which the government extended debt instruments to poor homeowners stands in stark contrast to how difficult it was for that same group to receive disaster relief aid.
5. Property damages and fatalities tell two different stories about the response to hurricane damages. It underscores the importance of directing disaster aid to effectively protect people and communities rather than prioritizing property. One way to do so could be by using public health measures, such as morbidity and mortality, as inputs into aid disbursement strategy decisions.

Implications for Public Health Practice

Public health practice includes healthy policy design. Due to the intimate connections between poverty, economic inequality, and health outcomes, any assessment of a policy’s impact on health should incorporate an assessment of its impact on economic equality and other social determinants of health. “Health in All Policies (HiAP) is a public policy approach that aims to increase the recognition of health and action on health, health determinants, and health equity across all sectors” (Ståhl & Koivusalo, 2020). HiAP is an active acknowledgement of the power of the social determinants of health and the impact of policy change on them. This study and others have identified socio-political responses to natural disasters as a strong social determinant of population health. Relationships between economic inequality and the rate of COVID-19 infections should lead us to extend HiAP to HEiAP, “Health Equity in All Policies.” It is urgent that in preparation for future public health crises, policy makers understand how inequitable federal aid distribution can deepen poverty in places that were poor prior to a disaster. Improved understanding should lead to further investigation and policy revision. HiAP cannot be limited to applying the same strategies everywhere. Legal scholar john a. powell (2008³¹) has developed a framework he refers to as Targeted Universalism: Equity 2.0. It is the marriage of a universal goal with targeted strategies to meet the goal. For example, the correlation between COVID-19 cases and economic inequality tells us that it is not enough to simply direct aid toward the areas where the most people are hurt. Aid distribution strategies should be intentionally targeted toward those most in need at every level from municipality to community to neighborhood on down to the household level. Targeting aid that way will help us advance toward two universal goals—reversing poverty growth and slowing the spread of diseases. Natural disasters and infectious disease epidemics have the potential to unite people against an outside (non-human) enemy and create solidarity. In this new context calls for equity in direct relationship to health would likely be well-received.

Disaster recovery experts and public health experts should collaborate with policy makers, community leaders, and social workers to define and implement a “Targeted Universalism” strategy encompassing preparedness, mitigation, adaptation, response, recovery, and resilience that defines universal goals and targeted strategies to achieve them. If communicated well (see below), such a strategy would improve the health and resilience of us all, because while COVID-19 does not discriminate and Hurricane María did not discriminate, both followed the longstanding patterns of compounded inequities that intensify harm and increase the vulnerability of everyone.

Much of public health practice is communications (Schaff and Dorfman, 2019). Messages about the causes and dangers to public health of widespread economic inequality is essential to protecting population health in Puerto Rico and elsewhere. It is essential to help the general public understand that property-focused disaster aid leads to more poverty, increased poverty leads to poorer health, and poor health in any part of the population helps infectious diseases to spread throughout the entire population. Our study illustrates these cycles.

There are lessons to be learned from the case studies about pandemic preparedness. Participants described the durability of social networks and how efficiently they mobilized after the hurricanes and in the early days of the pandemic. Others have noted that there was a high level of compliance in the Puerto Rican population when strict restrictions were put in place even before the first COVID-19 case was confirmed (Acevedo, 2021³²). Our finding that some post-hurricane response networks were still active or easily re-activated at the start of the pandemic highlights the importance of continuous engagement with the population. Public health practitioners and medical professionals should work closely with community members in prevention, surveillance, information and treatment efforts. According to Rodríguez-Díaz (2018³³), “The community response in Puerto Rico evidenced fundamental collective competencies that public health workers must nourish.” Health readiness could be improved by actively maintaining and regularly testing citizen networks. This could be done through community mapping to identify disaster vulnerabilities, through annual vaccine campaigns, or other activities that strengthen social bonds while improving disaster preparedness.

Dissemination of Findings

Five products are being developed to disseminate our findings: (1) a report of preliminary findings for the general public, (2) a list of implications for policies and programs, (3) recommendations for future federal aid disbursement that emphasizes economic equality, fosters resilience, and protects public health, (4) an informational website that features a map of Puerto Rico with visual representations of the data correlations, access to presentations, and other communication products created (including this report), and (5) public comment to FEMA contributing evidence-based recommendations on climate change, equity, and vulnerable populations. Town hall meetings will be held at both field sites to share findings and to solicit feedback from all participants. A two-hour seminar for the nonprofit sector, community leaders, graduate students studying public policy and social work, and public officials will be hosted in collaboration with the Departments of Social Work and Political Science of the Interamerican University in September, 2021. This event will be used to disseminate the findings to communities in Puerto Rico and discuss potential local applications.

Limitations

The primary limitations of the study result from the short research period demanded by quick response research. We were able to collect perishable primary data, but not all the secondary data we had hoped to include were available or accessible in a timely manner. This limited the quantitative analyses that we could conduct. Our study used only aggregated municipality-level data rather than individual household data, so we were unable to match measures at the individual level for Aims 1–3. The exclusion of crop damages from our data set is another important limitation given the correlation between agriculture and municipal poverty in Puerto Rico. Lastly, we included only federal public aid in our analyses. Estimates putting private aid amounts at $375 million are likely to be an undercount (Petrovich, 2018). By excluding non-public aid from our analyses we neglected to measure and incorporate the impact those dollars may have had.

Future Research Directions

Due to anthropogenic environmental changes, both infectious diseases and extreme weather events are expected to increase and intensify over the coming years (Preston, 2013³⁴). Understanding the health and economic impacts of different types of organizations in disaster recovery can guide policy-making and resource allocation decisions to better protect individual and community well-being. We referred above to the Smiley et al. (2018) study of organizational ecosystems and the impacts of changes in the number of different types of organizations in a post-disaster context, where they found that increased numbers of non-profit social and faith-based organizations were correlated with increased poverty growth, while increased numbers of local businesses and non-profit advocacy organizations were correlated with decreased poverty growth. These relationships were identified across all U.S. counties for the years 1999-2013, but that study did not include U.S. territories. It is urgent to identify whether the organizational ecosystem changes experienced in Puerto Rico are having similar impacts, because the nonprofit sector has roughly tripled in size since hurricanes Irma and María in 2017 (Estudios Técnicos, 2021³⁵). It is important to understand how (or if) growth in the nonprofit sector has affected poverty in Puerto Rico.

Future research should investigate similar questions related to private aid alone and to all public and private aid combined to determine their impacts and learn from their different strategies. Crop damages should be included in future research to reduce urban bias in property damage totals. Given the small proportion of the aid allocated to Puerto Rico that has arrived in the territory, and the even smaller amount of aid that the 78 municipalities have received, the effects of aid will only be fully detectable after more of the aid money is disbursed. The impacts of federal aid sent to Puerto Rico should continue to be investigated after all allocated funds have been disbursed.

Poverty increases vulnerability to hazard fatalities and infectious diseases. A detailed understanding of the economic landscape, and particularly its most economically vulnerable residents, is essential for Puerto Rico’s disaster readiness. Further exploration of relationships between economically-motivated and disaster-induced migration (to, from, and within Puerto Rico), and the impacts of population changes on poverty and disaster readiness, could usefully inform both disease prevention and hazard response strategies. Finally, it is urgent to understand how to reverse the pattern of recovery efforts leading to increased poverty. Finding and employing strategies that increase economic equality and reduce poverty will be essential to create better disaster responses in the future. Solving this problem becomes even more urgent as Puerto Rico recovers from the widespread effects of the COVID-19 pandemic. This will require a robust program of convergent research and should include scholars from the fields of communications and psychology, among others. Advancing this line of study will be a step toward moving public opinion away from the blaming focus which Belle (2006³⁶) described in the United States after Hurricane Katrina and which continues to pervade public narratives. “Motivated reasoning to justify existing economic and political systems and to maintain belief in a just world leads many… to explain economic inequalities as the result of purely individual factors.” As awareness of the health impacts of systemic racism continues to grow in the United States, explaining the role of institutions and natural hazards in poverty may drive the point home that poverty, and the vulnerability to health threats that go along with it, are not individual problems. Future research should focus on the entire disaster cycle beginning with mitigation and readiness and leading into response and recovery. Further research on the impacts of economic inequality on disaster readiness is imperative, and it must include pandemic readiness in addition to meteorological and geological disaster readiness. Research that guides movement beyond resilience to systemic change is urgently needed.

Endnotes

Endnote 1: We use the term Latine to be gender-inclusive and in recognition that there are more than two genders. There is great diversity in Puerto Rico in adoption of such terms, but in contrast to Latin@ and Latinx, the term Latine is used verbally and is easily pronounceable for native Spanish speakers (pronounced Lah-tee-nay). ↩

Endnote 2: Bridging social capital refers to the social bonds developed between different groups while bonding social capital refers to strengthening of social bonds within existing groups and linking social capital refers to connections made to institutions and decision-makers, such as those made by advocacy organizations. ↩

Endnote 3: The field research team was designed to include some residents of the communities studied. We wish to acknowledge the assistance of our research associates and site consultants: Juneilis Mulero Ortiz, Nicole Pecci Segrí, Gerardo Rivera, Lorena Bonilla, Anohiska Cardona, Paola Sánchez, Luis La Santa, Ana Hilda Rodríguez, and Daisy Vázquez.↩

Endnote 4: Sociologists Dr. Linda Colón, expert on poverty in PR, and Dr. Junia Howell, expert on inequality and natural hazards.↩

Endnote 5: Our data is a complete count of Puerto Rico municipalities, not a random sample. Thus, we do not use inferential statistics because they have no interpretable meaning. We standardized all variables on a scale to compare significance between them.↩

References