Disparities and Determinants of Food Insecurity for New Mexico’s Unique Populations

New Mexico, a majority-minority state, had the second-highest poverty rate in the USA (19.5%) and a highest rate of food insecurity (16.8%) in 2019. Its unique history means that nearly half of Hispanics in the state are from families that can trace their ancestral arrival back to the late 1500 s, an understudied subgroup called Hispaños (old New Mexican families). I use the Current Population Survey Food Security Supplement for the state of NM from 2000 to 2018 to look at food insecurity by Hispanic origin and nativity, attempting to isolate Hispaño households from other Hispanic origin households. Using linear probability models, results show that foreign- and native-born Hispanics were more likely to experience food insecurity than non-Hispanic New Mexicans of any nativity. Within those groups, Hispaños are less likely than their Mexican-origin peers to be food insecure. These results emphasize the importance of data disaggregation, and the results show promising results for further investigation into the outcomes of Hispanics with long-standing ties to an area versus those who are more recent arrivals.


Introduction
Food insecurity continues to be a concern for many US households.The US Department of Agriculture (USDA) defines food insecurity as a lack of "access by all people at all times to enough food for an active, healthy life."Households identifying as Hispanic have elevated rates of food insecurity, 16.2% compared to 11.1% in the US population overall.However, this broad categorization masks important diversity in lived experience and cultural history.For example, the descendants of the Spanish settlers in the Southwest, often referred to as Hispaños or Nuevomexicanos, have lived in what is now the USA for many generations with a distinct history and community.Hispaños represent 22% of the New Mexican population and 45% of the New Mexican Hispanic population.However, little is known about food insecurity levels or determinants for this group.This paper investigates food insecurity across Hispanic origin groups in NM, identifying the food insecurity rate for Hispaños and other Hispanics in the state and examining determinants of food insecurity.

Background
According to the USDA Economic Research Service, more than one in ten US households were food insecure in 2018, a number down from a high of 14.9% in 2011, after the Great Recession (Coleman-Jensen et al. 2019).A total of 4.3% of those households were classified as having "very low" food security, meaning that some household members ate less than they thought they should have, and household eating patterns (e.g., types of food) were disrupted.Food insecurity is highest among households near or below the federal poverty line, households with children (and especially those headed by a single parent), Black and Hispanic households, and those composed of only one person.Only about half of food insecure households reported participating in a federal food assistance program such as the Supplemental Nutrition Assistance Program (SNAP), Women, Infants, and Children (WIC), or the National School Lunch Program (NSLP).The lack of consistent access to food costs the USA about $170 billion in 2010 (Berkowitz et al. 2019) through the direct costs of addressing food insecurity as well as lower productivity and worse health caused by hunger.
Food insecurity is associated with a plethora of negative health and educational outcomes.For children, lower food security is associated with worse self-reported health, asthma, depressive symptoms, foregone medical care, and higher emergency department usage (Thomas et al. 2019).worse social-emotional outcomes among peers (Johnson and Markowitz 2018), worse school performance (Howard 2011;Johnson and Markowitz 2018), worse oral health (Chi et al. 2014), suicidal ideation (McIntyre et al. 2013), and behavioral issues, especially for boys (Whitaker et al. 2006;Jackson and Vaughn 2017).First-year college students experiencing food insecurity earn lower grades and may be less likely than food-secure peers to continue in higher education after the first year (Maroto et al. 2015;van Woerden et al. 2019).For adults, food insecurity is associated with higher rates of mental health issues (Whitaker et al. 2006;Cook et al. 2013;Martin et al. 2016) and diabetes (Mayer et al. 2015;Walker et al. 2019;Seligman et al. 2007).Food insecurity is associated with increased healthcare costs for older adults with chronic conditions, specifically stroke, arthritis, and diabetes, when compared to their food-secure peers (Garcia et al. 2018).Food-insecure older adults consume lower amounts of key nutrient categories than their peers and are more likely to have congestive heart failure, asthma, depression, poor gum health, have a hard time with day-to-day activities, or have had a heart attack (Ziliak and Gundersen 2017).
A highly discussed health outcome related to food insecurity is obesity, but the relationship is complex.Some sources show an association, but mechanisms are not agreed upon.There is a well-documented positive correlation between obesity and food insecurity for women in high-income countries, especially white and Hispanic women (Hernandez et al. 2017).One analysis tested and concluded that humans' instinct to stock up on fat stores to weather hard times may play a small part (Nettle et al. 2017).Another suggests a social-biological mechanism in which low socioeconomic status individuals may be predispositioned to both weight gain and food insecurity (Dhurandhar 2016).Still another study tests availability of healthy foods, time to grocery store, stress, and health insurance coverage as potential mechanisms but finds no evidence in favor of these (Kowaleski-Jones et al. 2019).Causal inference is lacking in the area.A dated but important longitudinal study on urban women found no association between changes in food insecurity and changes in weight (Whitaker and Sarin 2007), though another found that women in households that become food insecure are likely to gain weight (the effects for men were not as strong) (Wilde and Peterman 2006).
While the national average for food insecurity from 2016 to 2018 was 11.7%, it was significantly higher in NM at 16.8% (Coleman-Jensen et al. 2019).In 2018, 18.8% of New Mexicans had annual earnings below the poverty line, compared to the national average of 13.1% (US Census Bureau 2019a), and research shows that poverty is a major predictor of food insecurity (Chi et al. 2014).When eligible for SNAP or WIC, households of color are less likely to apply (Winham and Armstrong Florian 2015), and, even when they do and are approved, these programs do not cover the full cost of food for most families (Hamrick and Andrews 2016;Poblacion et al. 2017).
Racial and ethnic minorities are less likely to be food secure than their white non-Hispanic peers (Odoms-Young 2018).Latinx households are more likely to experience food insecurity than non-Hispanic white US households-16.2% of Hispanic households were food insecure in 2018 (Coleman-Jensen et al. 2019), and past economic instability has had varying effects for households of different races and ethnicities (Flores-Lagunes et al. 2018).This is especially relevant for NM, where, in 2018, Hispanics made up 49% of the population.Non-Hispanic white New Mexicans only account for 37.1% of the state population (U.S. Census Bureau 2019a).
The reasons for these disparities are many.First, families who struggle to obtain adequate foods are more likely to be poor (Chi et al. 2014), and New Mexicans face a 43.5% higher poverty rate than the national average (US Census Bureau 2019a, b, c).Poverty is a major predictor of food insecurity because low-income households are likely to face tradeoffs in purchasing necessities.When eligible for SNAP or WIC, households of color are less likely to apply (Winham and Armstrong Florian 2015), and even when they do and are approved, these programs do not cover the full cost of food for most families (Hamrick and Andrews 2016;Poblacion et al. 2017).These factors make clear the importance of examining food insecurity across population groups with and without controls; systemic inequities in access to education may cause the impact of race and ethnicity to load onto education coefficients in a regression setting, for example.
Spatial heterogeneity plays a nuanced part in disparate food security rates.Poor Latinx neighborhoods have fewer supermarkets than higher-income white neighborhoods (Moore and Diez Roux 2006;Raja et al. 2008;Walker et al. 2010) and lower access to quality food sources than socioeconomically similar white neighborhoods (Gordon et al. 2011).However, recent research suggests that food deserts only explain a small portion of food insecurity and that much of the disparity may lie on the demand side (Allcott et al. 2019;Zhen 2021), though it is important to note that relative prices, local economic effects of grocery store openings/closings (Adam and Jensen 2016), preferences based on previous location or income, and cultural preferences may affect demand heterogeneously for different groups (Karpyn et al. 2019).Page-Reeves and coauthors noted that, in conversations with community members in a historic Hispanic neighborhood and food desert in NM, many stated that the distance to the grocery store was a barrier to accessing enough healthy food (2014).A 2015 paper analyzed the effect that a new, subsidized grocery store had on a largely Hispanic food desert in the Bronx, NY, and found no effect on food security (Elbel et al. 2015).Some have pointed out that "food swamps," or areas with a high relative density of stores offering high-calorie or "junk" food to healthy foods, are better predictors of food insecurity than food deserts (Cooksey-Stowers et al. 2017;Rose et al. 2010).
Immigrants to the USA are much more likely to be food insecure than those born in the USA (Capps et al. 2002;Rabbitt et al. 2016;Flores-Lagunes et al. 2018).Nativity is a strong predictor of SNAP usage among Hispanics (Winham and Armstrong Florian 2015), and food security varies by Hispanic origin and immigration status (Rabbitt et al. 2016;Walsemann et al. 2017).It takes time for immigrants to build the social capital that can help them learn where to find low-cost food or to understand how to navigate the infamously complex US food and income assistance system.Furthermore, undocumented immigrants are ineligible for most federal food assistance so they must rely on food pantries or religious organizations in times of need.Language barriers and fear of immigration enforcement may create barriers to accessing an adequate food supply.The use of the broad Hispanic label, which typically combines immigrants and long-standing US residents, fails to account for these important factors.

New Mexico
New Mexico has a unique cultural history.The indigenous peoples-Navajo, Comanche, Apache, Ute, and Puebloof the area now known as NM were the sole inhabitants until the Spanish arrived in 1598.The Spanish claimed land and forcibly colonized the native peoples.Despite tensions, the indigenous peoples and the Spanish shared knowledge (water management systems, building techniques), traded (seeds, horses, livestock), and intermarried.The descendants of the Spanish settlers are often referred to as Hispaños or Nuevomexicanos."Anglos" (people of non-Iberian European descent) did not begin to settle in NM in significant numbers until around the time of the Treaty of Guadalupe Hidalgo in 1848.Note that many Hispaños are "mestizos"-that is, they are of both Native American and Hispanic descent, though, for the purposes of this analysis, the Native American and Hispanic categories are mutually exclusive.Thus, the state has a history fraught with racial and ethnic hierarchies, colonialism, and physical and cultural isolation predating the arrival of the railroad in the West (Salaz Marquez 1999).
The modern term Hispanic includes recent migrants to the area from Latin America and their descendants, as well as Hispaños whose families trace their arrival to the area to the Spanish settlers in the late sixteenth century.A total of 44.78% of New Mexican residents in the CPS sample for 2000-2019 identify as Hispanic.A total of 19.24% of New Mexican residents and 43.19% of Hispanic residents report their origin as "other Hispanic," which is used here as a proxy for Hispaño origin.For New Mexicans surveyed by Hunley and coauthors (2017), 92% of those who identified as "Spanish" also reported belonging to an old New Mexican family, which supports using "other Hispanic" as a stand-in for Hispaño identity, though only in NM.A total of 53.52% of Hispanic New Mexicans are US born Hispanics of Mexican origin and 3.28% are US-born Hispanics of Puerto Rican, Cuban, or Central or South American origin.Only 20.39% of Hispanics in NM are foreign born in the CPS sample.Distinguishing Hispanic origin subgroups and assessing food insecurity for Hispaños in particular offer a more accurate picture of ethnic differences in food insecurity.
There have been no comprehensive studies of the determinants of food insecurity in NM's Hispaño population to date.A qualitative study conducted by University of New Mexico researchers in collaboration with a community organization studied a small neighborhood near downtown Albuquerque.The 2300-member neighborhood has a strong Hispaño identity with many families tracing their lineage back to Spanish settlers-nearly 70% of residents are Hispanic or Latino, with only 17% of those identifying as immigrants (Page-Reeves et al. 2014).The neighborhood has high rates of poverty and is classified as an urban food desert, with 71% of residents reporting a need to travel to other neighborhoods to purchase food.Survey participants cited rising costs of food, stagnating wages, lack of jobs, distance/access to grocery stores, lack of literacy around food systems, and social stigma around accessing food assistance as barriers to food security (Page-Reeves et al. 2014).
This paper focuses on quantifying disparities in food insecurity by nativity and Hispanic origin using a representative sample of New Mexicans.It is the first quantitative analysis to examine differences in food security between Hispaños and Hispanics of Mexican origin, contributing to the small body of literature on Hispaños and improving our understanding of health disparities by disaggregating a vastly heterogeneous group of Hispanic identities and origins.

Data
This study uses the Current Population Survey (US Census Bureau 2019c) Food Security Supplement (USDA Economic Research Service 2021), abbreviated CPS-FSS, from 2000 to 2018 to examine food insecurity in NM.The CPS is a nationally representative survey that collects information on about 60,000 US households each month.The FSS is an additional set of questions asked to a subset of the regular December CPS respondent pool.Every household below 185% of the federal poverty level (FPL) is asked the full supplement as are households earning over 185% FPL, who indicate difficulty obtaining food in two screening questions.All other respondents are coded as food secure.The FSS then asks each household 18 questions relating to its food security.The responses are used to create an index of food security of adults and children in the household as well as the household as a whole.
The dependent variables used in this analysis are the CPS-FSS-constructed food security index variables, indicating household food security, and food security of adults and children separately within the household.These three outcomes were originally coded as a three-tiered categorical variable-food secure, low food security, and very low food security-but as is common in the literature using the CPS-FSS, I have dichotomized them into binary versions where household, adult, and child food insecurity are coded as 0 for "food secure" and 1 for "food insecure" if the household reports low or very low food insecurity.Race (white, Native American, other or mixed race, Hispanic/Latinx), nativity (native born, foreign born), and Hispanic origin (Mexican, Puerto Rican, Cuban, Central or South American, other Hispanic origin) are also captured in the CPS.Black and Asian households had very low representation in NM during the study period (n < 550 for both subgroups), so they were combined into an "other or mixed" race category for the purposes of this study.
I also control for other factors known to be predictive of food insecurity.Other CPS-FSS controls used are number of children in the household (continuous), female-headed household (binary), age of household head (0-17, 18-24, 25-44, 45-64, 65, and older), marital status (single, married or cohabitating, separated, divorced, or widowed), highest education level among household members (less than high school, high school, some college, 4-year college degree, or more), whether the household is above 185% of the FPL (binary), whether the head works for wages (binary), and whether the household received food stamps in the last year (binary)-all derived either from the FSS directly or the CPS basic monthly data from December of the relevant survey year.
The 2000-2018 CPS-FSS sample consists of 9616 New Mexican households comprised of 23,816 individuals, after dropping households, which did not completely answer the FSS (5800 respondents, or 19.6% of the full sample) and those with missing values for control variables (32 respondents, reflecting 0.11% of the original NM sample).Adult and child food securities were only measured separately from household food security from 2005 onward, yielding a slightly smaller sample size for those outcomes.

Statistical Analysis
I used linear probability models to examine the determinants of food security using the three binary outcomes described above-household food security, adult food security, and child food security.Ordered logistic marginal effects were similar in significance and magnitude and are not reported.I ran two main models: a model where the variable of interest is race/ethnicity interacted with nativity, and then one looking at Hispanic origin.The controls described above are included in all models presented below, except for Table 2, where controls are added in groups to each subsequent model.I use year-fixed effects to account for secular trends in food insecurity over time.Standard errors are robust and clustered at the household level, and the sample is weighted to account for the CPS sampling method.

Results
Table 1 below shows summary statistics for the weighted sample of individuals surveyed in the CPS-FSS from 2000 to 2018.Foreign-born Hispanics have higher rates of food insecurity than non-Hispanic New Mexicans or those of any other Hispanic origin who were born in the USA, and native-born Hispaños have statistically significantly lower unadjusted rates of food insecurity than their native-born Mexican origin peers.Figures 1 and 2 visually confirm what Table 1 shows: trends of food insecurity in NM from the CPS-FSS data.Hispanics tend to have higher rates of food insecurity than non-Hispanic New Mexicans, and foreignborn Hispanics seem to drive this effect.

By Nativity
In an extension to Myers and Painter (2017), I next analyzed the models with race/ethnicity and nativity variables (interacted manually in the creation of the variables), adding in covariates sequentially in columns.These LPM regression results are reported in Table 2 below.In the base model without controls, foreign-born Hispanics of any origin are 16.65 percentage points more likely to live in food insecure households than the base group, native-and foreign-born non-Hispanics.Column (1) also shows that US-born Hispanics of Mexican origin and of "other Hispanic origin" (likely Hispaño) are 6.89 percentage points and 4.51 percentage points more likely to be food insecure than their non-Hispanic counterparts, respectively.As controls are added in columns (2), (3), and (4), US-born Hispanics continue to be statistically significantly more likely to experience food insecurity (5.37 percentage points) than non-Hispanic households, but native-born Hispanic households show no statistical significance, indicating that, generally, nativity is associated with better food security for Hispanic households.

By Poverty Status
Table 3 looks at child and adult food insecurity in addition to overall household insecurity status.The left panel shows results for the full sample, and the right panel is restricted to the sample of households earning below 185% of the Federal Poverty Line.The full sample panel shows that foreign-born Hispanic adults and children are more likely to be food insecure than their non-Hispanic counterparts (9.52 and 4.8 percentage points, respectively), as are Hispaño adults (2.87 percentage points).I see similar results when I restrict the sample to low-income households, with low-income foreign-born Hispanic adults and children being more likely to be food insecure than the low-income base group (9.14 and 8.05 percentage points, respectively), along with Hispaño adults (6.71 percentage points).Since I am interested in heterogeneity between Mexicanorigin and "other" Hispanic-origin households in NM (Hispaños), I restrict the sample in the bottom panel of Table 3 to Hispanic households only, and I report LPM regression results where the base group for Hispanic origin is of Mexican origin.In the full sample, "other Hispanic origin"/Hispaño adults are between 1.5 pp (adults) and 3.3 (children) less likely to live in households where they are food insecure than their Mexican-origin peers, though the effect is not statistically significant.In the < 185% FPL subsample, Hispaños are between − 4.59 pp (child) and 1.48 pp (adult) as likely to be food insecure than the low-income base model, but, again, these estimates are not statistically significant.A large proportion of Hispanics of "other Spanish origin" are likely to be Hispaño, so this result suggests that my hypothesis that Hispaños fare slightly better than similar Mexican-identifying respondents may be correct, especially for the full sample of households.
Table 4 shows predictors of food insecurity in Hispaño households are age (seniors are less likely to be food insecure), low levels of education, female-headed households, lack of work, and poverty.Households who received SNAP in the last 12 months are more likely to report food insecurity, though note that this is endogenous-households are more likely to seek food assistance if they are having trouble affording food.

Conclusions
This paper adds to the growing body of work that emphasizes the importance of disaggregation in health and social science research, extending the examination of racial and ethnic heterogeneity in food insecurity to examine differences by Hispanic origin, as well as providing research on an understudied population: Hispaño New Mexicans.I find that foreign-born Hispanics are more likely than non-Hispanics to be food insecure, and that Hispaños are more food secure than their Mexican-origin counterparts.Immigrants face barriers to employment, education, food assistance, and often initially lack the social networks that could help them obtain those facilitators of food security.
New Mexico is one of the poorest and most food insecure US states.Though it has a small population, a plurality of New Mexicans are Hispanic.NM's unique history means that we have a group of Hispanics who have been in the area since 1598, plus more recent immigrants.The identity, cultural norms, population size in the USA, timing of and  reasons behind immigration waves, and racial and colonial history of a given subgroup can cause its members to have markedly different outcomes than those of other subgroups that share the same broad racial or ethnic identity.These factors make NM one of the most interesting places to study Hispanic disparities in access to food in the country.This paper is the first to examine food security among a particular understudied population, the approximately 45% of New Mexican Hispanics who identify as Hispaño (Hunley et al. 2017).This not only has implications for disaggregating food insecurity in NM but also shows the promise for evaluating the effect of skin color, language, culture, and immigration separately in other contexts.However, there are important limitations to consider when interpreting these results.First, these results are associational rather than causal, so it is important to note that the causal mechanisms behind the disparities in the findings may be many and interconnected, and more work is needed to disentangle the mechanisms from the outcomes here.Second, due to data constraints, I am unable to perfectly separate Hispaño families from other US-born Hispanics who do not identify as having Mexican, Puerto Rican, or Cuban ancestry.
This work adds to the growing literature showing that heterogeneity between Hispanic subgroups in terms of origin/ ancestry and immigration recency matters in terms of health behavior and outcomes including diabetes treatment (Urdaneta and Krehbiel 1989), risky sun behavior (Coups et al. 2012), cancer risk and outcomes (Stern et al. 2016), and, of course, food insecurity (Smith et al. 2016;Rabbitt et al. 2016).More in-depth examination of Hispanic heterogeneity and the questions that make up the food security indices, as well as differences within racial subgroups, is sorely needed.US Hispanics differ significantly in median age (Mexicans at 27 and Cubans at 40), poverty (Argentines at 9% and Hondurans at 26%), English proficiency (Hondurans at 48% and Spaniards at 93%, and citizenship (Puerto Ricans at 99%, Venezuelans at 51%) (Noe-Bustamante 2019), so it is conceivable that there are many more differences between groups.Hispaños are a useful comparison group because they, like all Hispanics, have been racialized as part of anti-Black and anti-Indigenous sentiment and policy in the USA but do not deal with issues related to recent immigration.Federal surveys do not ask about Hispaño identity, though some local surveys in NM do.Because most Hispaño-identifying people live in NM (Hunley et al. 2017) and only 1% of US Hispanics are of Spanish heritage (US Census Bureau 2019b), state-level surveys with a Hispaño indicator may be sufficient to determine food insecurity and health patterns for this understudied Hispanic subgroup.These differences indicate that broader policy prescriptions geared toward simplifying and streamlining the immigration process, reducing systemic discrimination, promoting less segregated neighborhoods, reducing healthy food prices relative to unhealthy substitutes, simplifying application processes for federal food assistance programs, and allowing undocumented immigrants to access more of these programs may be more effective at addressing food insecurity than simply targeting all Hispanics.This includes identifying which subgroups to focus on in the context of where to facilitate grocery store openings, outreach about food assistance programs, primary care information, and benefit generosity, and has large implications for governments, nonprofits, farmer markets, and other organizations addressing food insecurity, as they can better target those in most need.For example, in NM, it makes more policy sense to set up food banks and SNAP application assistance workshops in areas with large amounts of immigrants or Mexican-Americans than it does to do the same in predominantly Hispaño neighborhoods.The movement toward greater disaggregation of racial and ethnic groups is critical in order to accurately describe and address disparities in food insecurity and other dimensions of health.

Fig. 2
Fig. 2 NM food insecurity trends by Hispanic origin : unadjusted differences in food security status by Hispanic ethnicity and birthplace Column (2): differences in food security status by Hispanic ethnicity and birthplace, adjusting for household reference person demographic characteristics Column (3): differences in food security status by Hispanic ethnicity and birthplace, adjusting for household reference person demographic characteristics and household poverty/labor variables Column (4): differences in food security status by Hispanic ethnicity and birthplace, adjusting for household reference person demographic characteristics, household poverty/labor variables, and SNAP receipt Cluster-robust standard errors at the household level in parentheses *** p < 0.01, ** p

Table 3
Linear probab-2018 model regression results of Hispanic origin on food security status for Hispanics in NM,CPS-FSS, 2000-2018