Rental Assistance Improves Food Security and Nutrition: An Analysis of National Survey Data

Abstract

The U.S. is experiencing a severe housing affordability crisis, resulting in households having to make difficult trade-offs between paying for a place to live and basic health necessities such as food. Rental assistance may mitigate these strains, improving food security and nutrition. However, only one in five eligible individuals receive assistance, with an average wait time of two years. Existing waitlists create a comparable control group, allowing us to examine the causal impact of improved housing access on health and well-being. This national quasi-experimental study utilizes linked NHANES-HUD data (1999–2016) to investigate the impacts of rental assistance on food security and nutrition using cross-sectional regression. Tenants with project-based assistance were less likely to experience food insecurity (B = −0.18, p = 0.02) and rent-assisted individuals consumed 0.23 more cups of daily fruits and vegetables compared the pseudo-waitlist group. These findings suggest that the current unmet need for rental assistance and resulting long waitlists have adverse health implications, including decreased food security and fruit and vegetable consumption.

The United States (U.S.) is experiencing a severe affordable rental housing crisis. This crisis stems from the combination of rising housing costs, stagnant minimum wages, and limited available housing units.¹ Roughly half of U.S. renters spend over 30% of their income toward rent and are considered rent burdened according to the definition by the U.S. Department of Housing and Urban Development (HUD).² Lack of affordable housing access contributes to housing insecurity, which is broadly defined as experiencing rent burden, homelessness, frequent forced moves or evictions, or overcrowding.³ Furthermore, lack of affordable housing options can force renters to settle for substandard or hazardous housing and neighborhood conditions.⁴ These multiple forms of housing insecurity have significant implications for health behaviors and outcomes, including food security and nutrition.^3,5–8

Housing insecurity may affect food security and nutrition through multiple pathways.. First, housing costs compete with food costs. When households experience financial strain, they often make difficult trade-offs between basic health necessities.^6,7,9 Qualitative research demonstrates that for many households “the rent eats first,” leading to limited budgeting for food and other expenses.^9,10 Indeed, a national food bank study found over half of their clients reported struggling to decide between paying for housing or food.¹¹ Second, a lack of housing options could force households into neighborhoods with limited access to healthy foods. Specifically, low-income neighborhoods are more likely to exist in food deserts and have limited supermarkets.¹² Third, a lack of affordable and stable housing leads to homelessness, doubling up, and poorly maintained housing conditions. In these situations, individuals have less control over their food environment and may face challenges storing and preparing food.^4,9

Indeed, existing evidence suggests that housing insecurity affects nutrition and food security, with implications for multiple health outcomes including obesity and type 2 diabetes.^13,14 One study found families experiencing rent burden have nearly three times the odds of food insecurity compared to those not rent burdened.¹³ Another study found those experiencing housing insecurity had 40% increased odds of being food insecure in the future compared to those with housing stability.¹⁴ These findings suggest that improving access to stable, safe, and affordable housing may help to alleviate food insecurity and improve nutrition.

Programs and policies that reduce housing insecurity may simultaneously address food security. One such policy is HUD rental assistance, which subsidizes rent to 30% of a household’s income to meet the definition of affordable housing. Receiving rental assistance is associated with a reduction in housing insecurity, including an improved sense of stability and self-reported housing quality.^4,15 The two main types of rental assistance include project- and voucher-based assistance. Project-based housing includes public housing developments which are operated by local housing authorities, as well as privately owned developments which are subsidized through grants to their owners. Voucher-based housing is provided in the form of Housing Choice Vouchers (HCVs), formerly called Section 8 Vouchers, which are issued to tenants to subsidize private market rent.¹⁶ Despite the overall increase in recipients of HUD rental assistance over the past few decades, rental assistance need has continued to outpace supply, with only 20% of eligible households receiving assistance in 2020, a drop from 24% in 2005.^16,17 This unmet need for rental assistance creates quasi-experimental conditions which can be utilized to estimate the impact of rental assistance on health behaviors and outcomes.^18–20

Thus, in this study, we examine how access to rental assistance affects food security and nutrition among low-income households. In doing so, we seek to understand the health costs of unmet rental assistance needs and to examine the potential to improve food and nutrition through expanded housing access. Our analyses utilize a unique data linkage between the National Health and Nutrition Examination Survey (NHANES) and HUD administrative housing records. We compare those who currently receive assistance to those who will begin to receive assistance in the next two years using survey questions about food access and a food diaries on fruit and vegetable consumption. We hypothesize that access to rental subsidies may improve food security and nutrition by freeing up household finances to be spent on healthy food or by providing tenants access to more resourced neighborhoods that they would otherwise not be able to afford. While it is possible that gaining access to rental assistance may require moves which disrupt food routines and impede food access, we believe these costs will be outweighed by the benefits of rental assistance.

Methods

Data Source

This study used data collected between January 1999 to December 2016 from NHANES that has been linked with HUD administrative rental assistance records.²¹ Collected by the National Center for Health Statistics (NCHS), NHANES is a cross-sectional, multistage survey designed to assess health and nutritional status. The HUD administrative data provide a longitudinal record of entry and exit dates for rental assistance and the type of assistance received (project- vs voucher-based). Utilizing these data helps mitigate the bias of self-report of rental assistance participation.²² Respondents are eligible for linkage if they consent to data linkage and provide their social security number, date of birth, and sex. NCHS created weights to account for both linkage eligibility and nonresponse to make estimates representative of the noninstitutionalized civilian U.S. population. To ensure confidentiality of the linked data set, data were accessed and analyses were completed at the Federal Statistical Research Data Center. This study was exempted from review and informed consent by the Penn State Institutional Review Board because this analysis uses secondary and deidentified data.

Measures

Rental Assistance

Data on rental assistance came from the HUD administrative rental assistance records. The analytical sample was then limited to two groups, those currently receiving assistance at the time of NHANES interview (rent-assisted group) and those who entered rental assistance within two years following their NHANES interview (pseudo-waitlist group). By limiting our sample in this way, we ensure all participants will eventually receive assistance and the main difference between the control and experimental group is the presence or absence of assistance at the time of the interview. This limits selection bias and confounding as these two groups are expected to be economically and demographically similar to each other (e.g. adverse life events, health problems, the experience of homelessness, economic disadvantage). The time frame of two years was used to account for the average length of HUD waiting lists.²³ This quasi-experimental method has been previously used in work researching rental assistance and HUD housing. ^18–20 Food Security

We assess food security using a measure developed by the United States Department of Agriculture and included in the NHANES survey.²⁴ This measure includes 10 questions about the following topics: worrying food would run out before getting money to buy more; food didn’t last before being able to buy more; not being able to afford balanced meals; cutting the size of or skipping meals; eating less than they should; being hungry but not eating; losing weight; adults not eating whole day because they did not have money for food. Ratings for the scale were based on the number of affirmative responses to the 10 questions. Per standard NHANES protocol, adults were considered to have full food security if they gave no affirmative responses, marginal food security with 1–2 affirmative responses, low food security with 3–5 affirmative responses, and very low food security with 6–10 affirmative responses.²⁴ Our study explores adult food security using this 4-point scale and by examining a dichotomous severe food insecurity variable, in which participants are severely food insecure if they were in the “very low food security” category.

Daily Fruit and Vegetable Intake

Dietary intake data were assessed by two 24-hr recalls administered by trained interviewers using the USDA Automated Multiple-Pass Method in the What We Eat in America (WWEIA) portion of NHANES. Total fruit and vegetable (FV) intakes were determined using the Food Patterns Equivalents Database, which provides the cup equivalent (CE) amount of fruit and/or vegetable in a food. In this analysis, the CE was assessed as a continuous variable for total FV intake. The analysis of FV data utilized a subset of the linked records used in the food security analysis due to a change in how NHANES collected the data for the WWEIA portion of the survey. These data were collected from January 2003 to December 2016.

Demographic and Geographic Variables

Covariates in these analyses were collected from NHANES and are meant to adjust for remaining sociodemographic and geographic differences between the rent-assisted and pseudo-waitlist groups. Covariates include age, sex, race/ethnicity, highest level of education, and family income-to-poverty ratio. These covariates are expected to be equal for current and pseudo-waitlist households but might differ slightly if there are geographic or temporal differences between the two groups, adjusting for them can increase the precision of estimates.¹⁸Analyses also controlled for U.S. state of residence to account for geographic differences in housing supply and housing agency priorities. Finally, the NHANES survey cycle (2-year periods between 1999–2016) was also adjusted to account for temporal changes in rental assistance, food security, and FV consumption.

Analytical Strategy

All statistical analyses were conducted using Stata, version 17 (StataCorp LLC), used weights created by the NCHS that adjust for linkage eligibility, and accounted for the complex design of NHANES. Statistical significance was established at 0.05. Linear regression was used to examine relationships of rental assistance (current assistance vs pseudo-waitlist) with food insecurity and FV consumption. Logistic regression was used to explore the relationship between rental assistance and severe food insecurity, and we present marginal effects. Finally, we ran a sensitivity analysis to investigate if these relationships varied by type of HUD housing program (project-based vs voucher) by including an interaction term in the regression models. All models also adjusted for age, sex, race/ethnicity, highest level of education, and family income-to-poverty ratio.

Results

Demographics

The linked data file included 36,039 linkage-eligible adults, 18 years and older, who were interviewed in NHANES from 1999 to 2016 and had no missing information on the covariates and outcomes. Descriptive characteristics of the sample stratified by rental assistance status can be found in Table 1 for both the Food Insecurity and the Fruit and Vegetable samples. The sample in this analysis contains individuals who are at higher poverty levels than the broader linkage-eligible NHANES sample; this is expected since they are receiving or will receive HUD Assistance.^19,25

Table 1.

Demographic Variables by HUD Status

	Food Insecurity Sample (a)			Fruit & Vegetable Sample (b)
	Currently Receiving n = 1565 (c)	Pseudo-Waitlist n = 632 (d)	p-value	Currently Receiving n = 1115 (c)	Pseudo-Waitlist n = 290 (d)	p-value
Rental Assistance Program			0.009			0.032
Housing Choice Vouchers	54.3%	43.8%		56.2%	48.1%
Project-based housing	45.7%	56.2%		43.8%	51.9%
Age groups			0.006			0.011
18–44 years	52.3%	62.7%		50.0%	61.0%
45–64 years	26.8%	21.8%		31.4%	26.1%
65+ years	20.9%	15.5%		18.7%	13.0%
Sex			0.423			0.154
Male	25.8%	28.5%		25.3%	31.7%
Female	74.2%	71.5%		74.7%	68.3%
Race/ethnicity			0.470			0.575
Non-Hispanic white	35.8%	41.0%		37.3%	34.1%
Non-Hispanic Black	42.3%	40.0%		40.9%	43.8%
Other	5.7%	4.0%		3.3%	Suppressed
Hispanic	16.3%	15.0%		18.5%	Suppressed
Education			0.751			0.099
Less than High School	38.5%	37.9%		39.5%	32.6%
High School	30.7%	33.0%		29.7%	34.2%
Some College	24.4%	24.1%		25.5%	30.3%
College & higher	6.4%	5.0%		5.3%	2.9%
Poverty Level			0.005			0.138
0 – <50%	22.2%	17.3%		22.2%	16.9%
50% – <100%	39.6%	32.5%		39.4%	37.6%
100% – <200%	28.0%	31.7%		28.6%	28.0%
>= 200%	5.8%	11.6%		6.1%	11.7%
Missing	4.4%	6.9%		3.8%	5.9%
Prevalence of Food Insecurity Status			0.060
None	48.0%	55.8%
Low	15.7%	11.7%
Moderate	19.0%	15.2%
Severe	17.3%	17.3%
Unadjusted Mean Fruit & Vegetable Daily Consumption in Cups				2.160	1.819	0.007

Note: All individuals in the sample received rental assistance at some point during the observation period. P values shown for X² values test of the difference between current recipients and individuals in the pseudowaitlist group.

^asample collected from 1999–2016 (n = 2,197)

^bsample collected from 2003–2016 (n = 1,405)

^cReceiving rental assistance at time of the National Health and Nutrition Survey (NHANES) interview.

^dNot receiving assistance at time of NHANES, but will enter assistance within two years of the interview, the mean length of Housing and Urban Development (HUD) rental assistance waitlists.

The sample used for the food insecurity analysis included 2,197 adults who were currently receiving rental assistance (N = 1,565) or were in the pseudo-waitlist group (N= 632). The groups were similar in sex, age race, and education. The rental assistance group had a higher percentage of voucher-holders (p = 0.009) and were in a lower poverty bracket compared to those in the pseudo-waitlist group (p = 0.005). Food insecurity status was similar in the two groups (p = 0.06).

The sample used for the FV analysis included 1,405 adults who were currently receiving rental assistance (N = 1,115) or were in the pseudo-waitlist group (N= 290). Demographic comparisons in the FV subset were similar to those in the larger food insecurity sample, with current recipients being more likely to be voucher holders (p=0.032) and older (p=0.011). However, the relationship with poverty level did not reach statistical significance in the FV subset (p = 0.14). Those currently receiving rental assistance had a larger number of cups of FV consumed than those in the pseudo-waitlist group (Rental Assistance = 2.16, Pseudo-waitlist = 1.82, p = 0.007).

Rental Assistance Impacts on Food Insecurity

Overall, there was not a statistically significant difference in food insecurity by rental assistance when controlling for race, age, sex, education, and poverty level (Table 2). However, in stratified analyses based on rent assistance type, project-based assistance demonstrated the likelihood of food insecurity declines by 0.18-points for currently assisted versus pseudo-waitlist individuals (B = −0.18 [CI 95%: −0.33, − 0.03], p = 0.02). In contrast, no significant relationship was found for adults in voucher-based housing (B = 0.07 [CI 95%: −0.16, 0.29], p = 0.56). When looking at the marginal effects regression of the binary outcome, severe food insecurity (Table 2), the likelihood of severe food insecurity declines by about 0.5-points for those in the current rental assistance group vs those in the pseudo-waitlist group (B= −0.05 [CI 95%: −0.09, − 0.01], p = 0.02). However, in stratified analysis, this relationship only persists for project-based housing (B = −0.06 [−0.12, 0.00], p = 0.04) and not for voucher-based (B = −0.02 [−0.10, 0.04], p = 0.45), and the interaction did not reach statistical significance.

Table 2.

Analyses of Food Security and Fruit and Vegetable Consumption by Rental Assistance and Program Category: Linked NHANES–HUD Data, United States, 1999–2016

	All Rental Assistance d	Project-Based Assistance e	Voucher-Based Assistance e
Food Insecurity a	−0.08 (−0.23, 0.07)	−0.16 (−0.30, −0.01) *	0.09 (−0.14, 0.32)
Severe Food Insecurity b	−0.05 (−0.09, −0.01) *	−0.06 (−0.12, 0.00) *	−0.02 (−0.10, 0.04)
Fruit and Vegetable Consumption (cup/day) c	0.23 (0.01, 0.44) *	0.23 (0.06, 0.39) *	0.20 (−0.10, 0.49)

Notes: Models food insecurity, severe food insecurity, and daily fruit and vegetable consumption by cup outcomes among adults over 18. All models adjust for the complex survey design of the National Health and Nutrition Survey (NHANES) and are weighted to reflect eligibility for linkage to Housing and Urban Development(HUD) data. All models are adjusted for using covariates of age, sex, race/ethnicity, highest level of education, and family income-to-poverty ratio.

Values are presented as beta coefficients and parentheses are 95% confidence intervals.

Source: Authors’ calculations using NHANES-HUD linkage 1999–2016 for food insecurity and 2003–2016 for fruit and vegetable consumption analyses.

^aFood Insecurity is modeled using linear regression. (all rental assistance: n = 2,197, project-based assistance: n = 1,075, voucher-based assistance: n = 998)

^bSevere Food Insecurity is modeled using logistic regression, with marginal effects shown. (all rental assistance: n = 2,197, project-based assistance: n = 1,075, voucher-based assistance: n = 998)

^cFruit and Vegetable Consumption is modeled using linear regression (all rental assistance: n = 1,405, project-based assistance: n = 678, voucher-based assistance: n = 727)

^dThe rental assistance models compare current assistance recipients with pseudo-waitlist individuals, who will move on to receive assistance within the next two years, the mean length of HUD waitlists. This model includes all recipients, regardless of if they ultimately receive project- or voucher-based assistance

^ethe project-based and voucher-based assistance models are limited to those who ultimately receive rental assistance through project- or voucher-based assistance, respectively

^*p < 0.05

Rental Assistance Impacts on Fruit and Vegetable Consumption

On average, rent-assisted individuals consumed 0.23 more cups of fruits and vegetables per day (CI 95%: 0.01, 0.426) than those in the pseudo-waitlist group, holding all other variables constant (Table 2). When broken down by rental assistance type, individuals living in project-based housing consumed on average 0.23 (CI 95%: 0.06, 0.39) more cups of FV than those in the pseudo-waitlist group, and individuals with voucher-based housing on average consumed 0.19 (CI 95%: −0.10, 0.49) cups of FV, although this finding did not reach statistical significance nor did the interaction between project- and voucher-based assistance.

Discussion

Our national study uses a quasi-experimental approach to investigate if obtaining rental assistance impacts food insecurity and the consumption of fruits and vegetables. Our findings demonstrate that participants receiving rental assistance through project-based housing are significantly less likely to experience general and severe food insecurity. Additionally, both project- and voucher-based housing were associated with increased fruit and vegetable consumption, although this was only statistically significant for project-based housing. These analyses suggest rental assistance represents a platform for investing in family well-being and promoting healthier dietary lifestyles among low-income families.

Our research upholds existing literature, which has demonstrated a positive effect of rental assistance on health outcomes and behaviors. Prior research has shown that compared to those on rental assistance waitlists, adults receiving rental assistance had better overall health, were less likely to have uncontrolled diabetes, and were less likely to postpone or decline healthcare due to cost.^18–20 In general, these studies found that project-based rental assistance has a stronger and more significant impact on health than voucher-based assistance. Our study found similar relationships, although when formally tested we did not find significant differences between project and voucher-based assistance in our models.

These suggested differences between project-based and voucher-based housing are worthy of future study, particularly given policy shifts towards voucher-based rental assistance. Between 1993 and 2016, the number of households receiving voucher-based assistance increased from 1.20 million to 2.30 million while the number of project-based housing units decreased from 2.86 million to 2.39 million.¹⁶ This transition to voucher-based housing was part of an effort to deconcentrate poverty by allowing tenants to move to neighborhoods with less racial segregation and more economic opportunities.²⁶

Voucher-based rental assistance theoretically affords more housing choice than project-based assistance and thus may offer additional health benefits.² Indeed, some research has found that moves from project-based public housing to voucher-based housing are associated with health improvements. In particular, the Moving to Opportunity Study found that individuals who were randomly assigned to transition from project- to voucher-based housing had a lower prevalence of diabetes, extreme obesity, physical limitations, and psychological distress compared to those who remained in project-based assistance.²⁷ However, additional literature has documented challenges related to voucher-based housing. Research has demonstrated that while residents who relocated using vouchers tend to live in neighborhoods with slightly lower poverty rates, they often ended up living in neighborhoods which are still poor and racially segregated.²⁶ These households also navigate additional challenges compared to those receiving project-based assistance including paying for utilities and expensive security deposits, having unforgiving rent timelines, and finding available units in a tight market.^28,29 These elements of voucher-based housing may undermine food security and nutrition.

In contrast, characteristics of project-based housing developments may promote nutrition and food security. For example, project-based housing may provide more resources to tenants by offering a central location for residents, allowing for easy access to programs which provide fruits and vegetables for pick-up at housing sites, on-site community gardens, and health courses, such as diabetes prevention and management classes.^30–32 Additionally, project-based housing residents may feel a closer sense of community, allowing for neighborhood social support and norms to play a larger impact on influencing healthy behaviors such as fruit and vegetable consumption.^33,34

Regardless of differences across programs, our findings suggest that both project- and voucher-based rental assistance are valuable policies that can improve access to food and nutrition. Housing insecurity, food insecurity, and nutrition have all been linked to adverse health outcomes, decreased access to medical care, and mental health challenges. In demonstrating how housing policy impacts health and nutrition, this study emphasizes that policymakers must consider the health effects of legislation in all domains, even those not obviously linked to health. As we see from this study and the growing body of literature demonstrating benefits of rental assistance, there are substantial health costs to waiting for assistance. With 80% of eligible households not receiving rental subsidies, it is critical for policy makers to increase funding for this resource.

Limitations

The current study has limitations that should be noted. First, data collected in WWEIA was based on recall in surveys, which can be inaccurate. However, this should not systematically be different between rent-assisted and pseudo-waitlist groups. Second, certain groups may be prioritized for rental assistance and be overrepresented in the current assistance group. But because they are mostly more disadvantaged groups such as families experiencing homelessness, pregnant women, or people with disabilities and chronic health conditions, this would tend to make our findings more conservative.^35,36 Third, our study was not able to account for other potential mediators of rental assistance with food insecurity and fruit and vegetable consumption, such as proximity to grocery stores, housing quality, or food stamp recipient status. Fourth, as an observational study, our study is vulnerable to remaining differences between pseudo-waitlist and assisted groups. However, by utilizing the pseudo-waitlist model, our quasi-experimental approach limits selection bias that could exist, such as motivation to enter into rental assistance and local availability of HUD assistance.

Future Research

This analysis contributes to growing literature on the health costs of unmet housing needs and has implications for population health equity. However, it does not address the racial disparities seen in food and housing security. Food insecurity and affordable housing challenges are unequally borne by communities of color owing to an ongoing history of discriminatory employment, incarceration, and housing policies and practices.^6,37,38 People of color are twice as likely to experience food insecurity compared to non-Hispanic whites, with the frequency of racial discrimination being a significant predictor of food insecurity.^38,39 Future research should investigate how race and racism may shape the health impacts of rental assistance.