Prevalence and risk factors of food insecurity among Californians during the COVID-19 pandemic: Disparities by immigration status and ethnicity

Abstract

The COVID-19 pandemic exacerbated socioeconomic disparities in food insecurity. Non-citizens, who do not qualify for most publicly-funded food assistance programs, may be most vulnerable to food insecurity during the pandemic. However, no study has examined heterogeneity in food insecurity by immigration status and ethnicity in the context of the pandemic. We analyzed the 2020 non-restricted California Health Interview Survey to examine disparities in food insecurity by ethnicity and immigration status (i.e., US-born, naturalized, non-citizen) among Asians and Latinxs (N = 19,514) compared to US-born Whites. Weighted multivariable logistic regression analyses assessed the association of immigration status and ethnicity with food insecurity. Decomposition analyses assessed the extent to which pandemic-related economic stressors, including experiencing reduced work hours or losing a job versus pre-pandemic socioeconomic position (SEP), accounted for disparities in food insecurity by ethnicity and immigration status. Regardless of immigration status, Latinxs were more likely to experience food insecurity than Whites. Based on the adjusted analyses, non-citizen, naturalized, and US-born Latinxs had a predicted probability of 12%, 11.4%, and 11.9% of experiencing food insecurity, respectively. In contrast, non-citizen Asians, but not US-born or naturalized Asians, reported greater food insecurity than Whites (12.5% vs. 8.2%). SEP accounted for 43% to 66% of the relationship between immigration status-ethnicity and food insecurity. The pandemic exacerbated economic hardship, but food insecurity was largely explained by long-standing SEP-related factors among Latinxs, regardless of immigration status, and non-citizen Asians. To address disparities in food insecurity, social assistance programs and COVID-19 economic relief should be extended to non-citizens.

1. Introduction

While food insecurity in the United States (US) has slowly declined since the 2008 Great Recession, the national rate of food insecurity peaked at a high of 23% of households between April and May 2020 (Schanzenbach and Pitts, 2020) and was estimated to include 42 million people in 2021 (Gross et al., 2021). Food insecurity is caused by multifaceted social and economic factors, such as low-income relative to the cost of living, limited wealth, or lack of social and financial resources to weather an economic shock to a household's budget, resulting in the inability to support the nutritional needs of one or more household members (Odoms-Young and Bruce, 2018; Gundersen and Ziliak, 2018). As a social determinant of health, food insecurity is associated with adult adverse health outcomes, including cardiovascular disease, metabolic syndrome, and depression (Gundersen and Ziliak, 2015; Lee et al., 2018; Bishop and Wang, 2018; Saiz et al., 2016), higher health care expenditures (Berkowitz et al., 2018), and low medication adherence (Knight et al., 2016). National estimates find that while the food insecurity rate remained relatively stable between 2019 and 2020 (10.5%) despite its early pandemic peak, Latinx (17.2%) and those unable to work during the COVID-19 pandemic (20.4%) were disproportionately affected (Coleman-Jensen et al., 2021). While no national-level data on immigrants' food insecurity rate exist, estimates range from 30 to 60% (Altman et al., 2020; Maynard et al., 2019). Given pre-existing social disparities and negative economic shocks due to the pandemic, immigrants and minoritized racial-ethnic groups may be at greater risk of food insecurity. Investigating the extent to which COVID-19-related economic stressors and pre-pandemic socioeconomic factors explain food insecurity among individuals of varying citizenship status and race-ethnicity can inform the development of more effective health interventions and policies to advance health equity.

The social responses to contain the COVID-19 pandemic exacerbated economic instability, a strong predictor of food insecurity (Gundersen and Ziliak, 2018). In particular, shelter-in-place and telework contributed to unemployment (Huang et al., 2016), decreased work hours (Shelton, 2022), and increased the official poverty rate (i.e., cash resources) to 11.4% in 2020 from 10.5% in 2019, translating to an additional 3.3 million people experiencing poverty (Pereira and Oliveira, 2020; Shrider et al., 2021). During the early stages of the pandemic in April 2020, unemployment rates reached an all-time high of 14.7%, the highest since the Great Depression (1929–1939), with minoritized racial-ethnic groups and immigrants experiencing the highest rates (Kochhar and Bennett, 2021). The federal government provided three Economic Income Payments and unemployment benefits to individuals with a valid legal status to compensate for lost income. COVID-19 relief programs assisted families in weathering the pandemic's economic effects and preventing food insecurity increases (Raifman et al., 2021). In response to the exclusion of individuals without a valid Social Security Number from federal pandemic aid, states had the option to use their funds to assist a limited number of non-citizens. For example, California provided a one-time stimulus check of $500 to 0.33% (150,000) of the state's estimated two million undocumented individuals (Governor Newsom Announces New Initiatives to Support California Workers Impacted by COVID-19, 2022; Hayes and Hill, 2017).

Given the de jure exclusion of undocumented individuals from the robust federal COVID-19 social safety net, there is heterogeneity in the economic impacts of the pandemic by citizenship status. Increases in the official poverty rate from 2019 to 2020 were more pronounced for non-citizens (16.3% to 17.8%, +2.5 percentage points) than naturalized citizens (9% to 9.2%, +0.2 percentage points) and US-born individuals (10.1% to 11.1%, +1 percentage point) (Shrider et al., 2021). Latinx individuals also experienced the most substantial increase (15.7% to 17%) in poverty among all racial-ethnic groups, while Asian individuals experienced the lowest (7.4% to 8.1%) (Shrider et al., 2021). Further, a report finds that 68.8% of Latinx adults in a household with at least one non-citizen experienced a job loss or reductions in their work hours or income compared to 49.1% of Latinx adults in which all family members are citizens and 38% of White adults (Gonzalez et al., 2020). Given that job loss is predictive of current and future food insecurity (Altman et al., 2022), and the largest socioeconomic effects during the pandemic were greatest among marginalized groups (Artiga and Rae, 2020; Perry et al., 2021), racially minoritized immigrants already living on the margin may experience the most profound risk of food insecurity during the COVID-19 pandemic.

While COVID-19 posed novel socioeconomic challenges, immigrants had already faced a myriad of social and economic risk factors of food insecurity. Before the pandemic, immigrants were more likely than the general population to experience adverse social conditions(Proctor et al., 2016) and fewer economic opportunities (De Trinidad Young et al., 2018). For instance, immigrants (12.5%) were already more likely to be in poverty compared to their US-born (10.1%) counterparts in 2019 (Shrider et al., 2021). Historically, non-citizen immigrants are overrepresented in low-wage industries and the informal labor market, characterized by job instability and limited employment-based benefits (Haro et al., 2020). While they are more likely to experience risk factors for food insecurity, the federal government has excluded most non-citizens from any public assistance programs since 1996 (Fox, 2016). Moreover, the 2020 Public Charge Rule played a leading role in discouraging immigrants from accessing health and food public benefits. Even if eligible, many immigrants feared their participation in public benefits would hurt their future chances of obtaining legal status or lead to deportation (Perreira et al., 2018; Caldwell et al., 2020). This chilling effect manifested in decreased participation in the Supplemental Nutrition Assistance Program (SNAP) among mixed-status families with US-born children (Barofsky et al., 2020). In combination with pre-existing exclusionary policies that shape socioeconomic opportunities, the pandemic may have worsened pre-existing drivers of food insecurity among already vulnerable immigrant groups (Page et al., 2020).

In the US, immigration status and ethnicity are axes of social stratification that have consequences for immigrants' and minoritized racial-ethnic groups' socioeconomic prospects (Menjívar, 2014). For example, Latinx comprise a larger share of the undocumented population and lacking a valid legal status limits access to public programs, stable jobs, and other economic resources (Walsemann et al., 2017). As such, immigration status may differentially shape ethnic groups' social and economic outcomes and, thereby, their risk for food insecurity during the pandemic. Previous work suggests that the experiences and effects of immigration status on food insecurity vary by race-ethnicity (Walsemann et al., 2017), and more recent work finds that citizenship status moderates the relationship between disability and food insecurity among adults (Altman et al., 2020). However, limited research has focused on the intersection of ethnicity and immigration status in the context of the COVID-19 pandemic. Recent studies examining the effects of the pandemic on food insufficiency (a proxy for food insecurity) have used the Census Household Pulse Survey, which does not allow for the disaggregation of respondents' immigration status or race-ethnicity (Altman et al., 2022).

Given immigrants' ethnic heterogeneity, we seek to better understand the prevalence of food insecurity among California's two largest immigrant groups—Latinxs and Asians—during the pandemic. In this study, we examine the extent to which disparities in food insecurity by immigration status-ethnicity exist. We also disentangle the variation in food insecurity explained by COVID-19 economic stressors and pre-pandemic socioeconomic position (SEP) to understand how these key factors shape disparities in food insecurity by immigration status-ethnicity. Because immigrants' experiences and socioeconomic prospects vary by citizenship status and race-ethnicity, we expect that non-citizen immigrants from minoritized racial-ethnic groups will have the greatest likelihood of food insecurity compared to their US-born White counterparts. We hypothesize that Latinx non-citizens will be particularly more likely to be food insecure than US-born Whites because they comprise California's largest share of the undocumented population, and non-citizens adults who have been lawful permanent residents for less than five years are barred from SNAP. Lastly, we expect that COVID-19 economic stressors will be a primary driver of food insecurity among non-citizens, as they experience the most precarious social and economic conditions.

2. Methods

2.1. Data

We use publicly available and anonymized data from the 2020 California Health Interview Survey (CHIS) of adults (age 18 and older). The CHIS is one of the most comprehensive state-level surveys in the US and provides rich and detailed information on California's health and social profiles. The 2020 CHIS was conducted via the Internet or computer-assisted telephone interviews in English, Spanish, Mandarin, Cantonese, Korean, Vietnamese, and Tagalog. The 2020 CHIS is also one of the few publicly available datasets to provide data on COVID-19 experiences and stressors (Ponce et al., 2021). A total of 21,949 people completed the CHIS from March to October 2020. We restrict our analysis to those who had complete data and were part of our race and ethnic groups of interest (i.e., non-Latinx White, Latinx, and Asian) for an analytical sample of 19,514 adult respondents. The study did not constitute human subjects research and was exempted from review.

2.2. Variables

Food insecurity was measured using the 6-item short-form of the US Department of Agriculture (USDA) Household Food Security Module. The 6-items are summed to create an overall score of food insecurity (range: 0–6). The USDA classifies a score of 2 or higher as food insecure, which we code as “1” and otherwise as “0.”

Our primary independent variable of interest was immigration status by ethnicity. We used self-reported immigration status (US-born, naturalized, and non-citizen) and ethnicity (non-Hispanic White, Latinx, and Asian). It is important to note that non-citizens include lawful permanent residents, undocumented and DACAmented persons, students, refugees, and temporary workers. Our final immigration status by race-ethnicity categories was as follows: US-born Latinxs, US-born Asians, US-born Whites, Naturalized Latinxs, Naturalized Asians, Non-citizen Latinxs, and Non-citizen Asians. Due to small sample sizes, we could not include other groups (e.g., naturalized Whites or non-citizen Black). We focused our analysis on the two largest groups, Latinxs and Asians, who comprise 89% of immigrants in California (Johnson et al., 2021).

2.3. Covariates

Our covariates were selected apriori based on previous research (Walsemann et al., 2017). We included several demographic, familial, and pre-pandemic socioeconomic position (SEP) covariates that could explain the association between immigration, ethnicity, and food insecurity. Our demographic covariates included age, marital status, and sex. Our familial covariates included household size, whether participants lived in a rural area, and family structure. Our two independent SEP indicators included (1) the highest level of educational attainment and (2) whether participants' pre-pandemic 2019 household income was below 100% of the Federal Poverty Limit (FPL).

Finally, we considered the effects of COVID-19 economic stress on food insecurity. We examined two economic stressors attributed to the COVID-19 pandemic: (1) experiencing reduced work hours or (2) losing a job. We examined whether participants experienced any of the two COVID-19 economic stressors, with “0” indicating that the participant did not experience any and “1” denoting experiencing at least one.

2.3.1. Statistical analysis

We first examined the distribution of food insecurity and demographic, familial, pre-pandemic SEP, and COVID-19 economic stressor covariates among the entire sample and by immigration status and ethnicity category. We use the chi-square test of association for categorical variables and t-tests for continuous variables. Next, we evaluated a series of nested logistic regression to examine the association between immigration status-ethnicity category and food insecurity. To account for the complex survey design used in CHIS, all analyses were weighted using jack-knife replication weights (80 total weights) (Sherr et al., 2021). Model 1 estimated the log odds of food insecurity by immigration status-ethnicity, adjusting for demographic and familial covariates. Model 2 added SEP indicators to assess whether educational attainment and pre-pandemic income explained differences in food insecurity by immigration status-ethnicity. Finally, Model 3 added the indicator of COVID-19 economic stressors. We use the margins command in STATA to calculate marginal predicted probabilities to aid in interpreting each model and estimate the relative percentage change in the predicted probabilities of experiencing food insecurity. We also formally test whether the predicted probabilities of food insecurity vary across immigration status-ethnicity categories compared to the reference group (US-born Whites) using the contrast command in STATA. We also present the marginal predicted probabilities of food insecurity based on Model 3.

We used the Karlson, Holm, and Breen (KHB) decomposition analysis to evaluate the extent to which COVID-19 economic stressors and pre-pandemic SEP explained the association between immigration status and ethnicity with food insecurity. KHB decomposition analysis is helpful because it accounts for the distribution of non-linear variables, including binary variables. KHB summarizes the direct, indirect, and total effects of the association of critical independent variables on outcomes. Finally, KHB also summarizes the percent mediation. Previous studies have used KHB to conduct decomposition analyses on CHIS data (Bacong and Sohn, 2021; Sohn and Bacong, 2021). All analyses and recoding were done using Stata Version 17.0.

2.3.2. Sensitivity analyses

We examined the role of employment status in the past week as an additional indicator for SEP. However, we excluded employment status as an indicator of SEP in our primary analysis because of potential endogeneity with COVID-19 economic stressors (i.e., lost jobs due to the pandemic). The results of these analyses are included in Appendix A.

3. Results

Table 1 presents sample characteristics by immigration status-ethnicity in 2020. US-born Whites comprised 63.8% of our analytic sample, 22.1% were Latinx, and 14.1% were Asian. The overall prevalence of food insecurity in our sample was 10.4%. Significant variation in food insecurity was present, with non-citizen Latinxs reporting the highest levels of food insecurity (23.5%) and US-born Asians reporting the lowest (4.4%). Compared to US-born Whites, a higher share of non-citizen Asians (46.9%), US-born Asians (56.6%), and Latinxs of all immigration statuses were between the ages of 18–34. On average, non-citizen, naturalized, and US-born Latinxs reported a larger household size than US-born Whites, with 4.2, 3.8, and 3.5 individuals per household, respectively. Further, a greater share of US-born Whites lives in rural areas (17.6%).

Table 1.

Sample characteristics by immigration status and ethnicity, California Health Interview Survey 2020 Adult Respondents, Weighted estimates, N = 19,514.

	US-born Whites	US-born Latinxs	US-born Asians	Naturalized Latinxs	Naturalized Asians	Non-citizen Latinxs	Non-citizen Asians	Total	p-value
	N = 12,452	N = 2769	N = 870	N = 1049	N = 1438	N = 499	N = 437	N = 19,514
Food insecure, %	4.9	14.9	4.4	14.1	4.8	23.5	13.5	10.4	<0.001
Demographics
Age categories, %									<0.001
18–34	22.2	53.6	56.6	14.9	11.1	26.1	46.9	31.0
35–49	20.6	24.6	23.3	26.4	27.2	42.4	29.6	25.2
50–64	25.4	12.5	10.6	36.5	33.9	25.9	18.8	23.3
65+	31.8	9.3	9.6	22.1	27.8	5.6	4.6	20.5
Marital status, %									<0.001
Married	52.8	35.7	34.9	64.6	70.1	57.9	57.4	51.1
Living w/ partner	9.8	11.6	8.9	7.0	2.2	16.6	4.1	9.8
Wid/Sep/Div	18.3	9.8	6.5	16.1	14.6	9.1	8.1	14.0
Never married	19.0	42.8	49.7	12.3	13.1	16.4	30.4	25.1
Female, %	51.1	49.8	56.0	51.0	54.4	51.2	45.5	51.0	<0.001
Household size, mean (SE)	2.59 (0.02)	3.52 (0.04)	3.06 (0.06)	3.76 (0.07)	3.29 (0.05)	4.15 (0.08)	3.31 (0.08)	3.19 (0.01)	<0.001
Rural, %	17.6	9.3	3.2	8.6	5.3	11.5	3.5	12.0	<0.001
Family structure, %									<0.001
Single HH, no children	43.4	46.5	59.5	27.0	27.8	24.8	36.3	39.9
Single HH with children	4.3	18.6	6.2	9.7	3.3	17.7	8.0	9.8
Dual HH, no children	37.1	16.4	18.9	38.3	43.4	22.9	23.2	30.0
Dual HH with children	15.2	18.5	15.3	25.0	25.5	34.7	32.5	20.3
Education, %									<0.001
Less than HS	3.7	10.8	2.6	43.9	17.9	55.0	16.5	16.1
HS graduate/GED	22.3	27.5	7.1	22.1	12.8	22.5	14.5	22.0
Some college	16.9	21.2	15.5	12.4	8.8	11.3	8.5	16.0
Associate's degree	6.2	8.4	4.5	3	4.3	1.0	2.2	5.5
Bachelor's degree and above	50.9	32.1	70.3	18.6	56.2	10.2	58.2	40.4
<100% FPL	6.4	18.2	8.7	20.1	9.8	29.6	20.6	13.9	<0.001
Any COVID-19 economic stressors	22.7	30.2	27.9	24.9	23.9	42.4	28.9	27.1	<0.001
Reduced hours due to COVID-19	16.5	20.9	20.1	16.8	16.0	29.3	17.7	18.9	<0.001
Lost job due to COVID-19	8.0	12.9	10.2	9.9	9.3	18.8	14.8	10.9	<0.001

Note: We use the chi-square test of association for categorical variables and t-tests for continuous variables.

Non-citizen Latinxs were also more disadvantaged than US-born Whites. Non-citizen Latinxs were more likely to have less than a high school diploma ( 55.0% vs. 3.7%), live in poverty (29.6% vs. 6.4%), experience reduced hours due to the COVID-19 pandemic (29.3% vs. 16.5%), and have lost their job during the pandemic (18.8% vs. 8.0%). While non-citizen Asians are more likely than US-born Whites to have a Bachelor's degree (58.2% vs. 50.9%), 20.6% live in poverty, and 14.8% lost their job due to COVID-19. Overall, 68.5% of our sample was US-born, 18.7% were naturalized, and 12.9% were non-citizens. Latinxs comprise a higher share of all non-citizens (62.5%) in our sample than Asians (26.4%).

Table 2 presents log-odds of food insecurity as a function of immigration status-ethnicity. After controlling for demographics (Model 1), US-born Latinxs (b = 0.93, SE = 0.11), naturalized Latinxs (b = 1.07, SE = 0.18), and non-citizen Latinxs (b = 1.57, SE = 0.05) were significantly more likely to be food insecure than US-born Whites. Among Asians, non-citizens (b = 1.04, SE = 0.23) were more likely to report food insecurity than US-born Whites. After adjusting for pre-pandemic SEP (Model 2), immigration status-ethnicity disparities in food insecurity were attenuated but remained. Latinxs of all immigration statuses and non-citizen Asians were more likely to be food insecure than US-born Whites even after accounting for SEP, suggesting other important factors may be driving food insecurity. Model 3 assessed whether COVID-19 economic stressors explained the disparities in food insecurity by immigration status-ethnicity. Except for the coefficient of non-citizen Latinxs (b = 0.52, SE = 0.20), the coefficients for all other immigration status-ethnic groups did not change in magnitude nor significance when COVID-19 economic stressors were considered. Notably, there were no differences in food insecurity between US-born and naturalized Asians and US-born Whites for any of the models.

Table 2.

Log-odds of food insecurity regressed on immigration status-ethnicity, California Health Interview Survey 2020 Adult Respondents, weighted analysis (N = 19,514).

	Model 1	Model 2	Model 3
	b (SE)	b (SE)	b (SE)
US-born Latinxs (ref. US-born Whites)	0.93***	0.51***	0.51***
	(0.11)	(0.12)	(0.12)
US-born Asians	−0.32	−0.27	−0.33
	(0.24)	(0.24)	(0.24)
Naturalized Latinxs	1.07***	0.42*	0.44*
	(0.18)	(0.19)	(0.19)
Naturalized Asians	0.03	−0.17	−0.19
	(0.18)	(0.19)	(0.19)
Non-citizen Latinxs	1.57***	0.60**	0.52*
	(0.15)	(0.20)	(0.20)
Non-citizen Asians	1.04***	0.59*	0.57*
	(0.23)	(0.23)	(0.24)
Age = 35–49 (ref. 18-34);	0.01	0.07	0.07
	(0.14)	(0.14)	(0.14)
Age = 50–64	−0.03	−0.21	−0.21
	(0.15)	(0.16)	(0.17)
Age = 65+	0.04	−0.37+	−0.23
	(0.19)	(0.20)	(0.21)
Cohabiting (ref. single)	0.24	0.21	0.17
	(0.32)	(0.30)	(0.29)
Wid/Div/Sep	0.33	0.12	0.12
	(0.34)	(0.32)	(0.31)
Never married	0.22	0.05	0.04
	(0.32)	(0.30)	(0.29)
Female (ref. male)	0.27**	0.12	0.14
	(0.09)	(0.10)	(0.10)
Household size	0.11**	0.03	0.03
	(0.03)	(0.04)	(0.04)
Rural (ref. urban)	0.01	−0.17	−0.15
	(0.12)	(0.13)	(0.13)
Family structure type (ref. single household, no children)	−0.71*	−0.63+	−0.61+
	(0.35)	(0.32)	(0.31)
Dual household, no children	−0.29	−0.30	−0.29
	(0.34)	(0.32)	(0.31)
Dual household with children	0.43**	0.06	0.07
	(0.15)	(0.17)	(0.17)
HS graduate/GED (ref. no HS)		−0.47**	−0.43**
		(0.14)	(0.14)
Some college		−0.44**	−0.45**
		(0.15)	(0.15)
Associate's degree		−0.56**	−0.57**
		(0.18)	(0.18)
Bachelor's degree and above		−1.57***	−1.55***
		(0.15)	(0.15)
≥ 100% FPL (ref. <100% FPL)		−1.84***	−1.87***
		(0.12)	(0.12)
Any COVID-19 economic stressors			0.57***
			(0.10)
Constant	−3.36***	−0.23	−0.47
	(0.35)	(0.40)	(0.39)

*** p < 0.001, ** p < 0.01, * p < 0.05, + p < 0.1; standard errors in parentheses.

Table 3 presents the predicted probabilities of food insecurity by immigration status-ethnicity for each model of Table 2. In Model 1, non-citizen Latinxs have the highest predicted probability (21.3%) of food insecurity net demographic factors. In Model 2, when we adjust for pre-pandemic SEP, we find a percent change decrease of 40.8%, 22.4%, 10.7%, and 7.1% in the predicted probabilities of food insecurity among non-citizen Latinxs (21.3% to 12.6%), naturalized Latinxs (14.3% to 11.1%), non-citizen Asians (14.0% to 12.5%), and US-born Latinxs (12.7% to 11.8%), respectively, compared to the base model. In Model 3, we adjust for COVID-19 economic stressors and find that only among non-citizen Latinxs did the predicted probability of insecurity slightly change (4.8% decrease; 12.6% to 12.0%). Unlike other groups, the predicted probabilities of food insecurity for US-born Whites and US-born Asians did not decline when we incrementally adjusted for SEP and COVID-19 economic stressors. Fig. 1 demonstrates the predicted probabilities of food insecurity from Model 3. Based on the post-estimation results, the Wald tests indicate the probabilities of food insecurity for US-born (p = 0.0002), naturalized (p = 0.0256), and non-citizen (p = 0.0355) Latinxs and non-citizen Asians (p = 0.0296) were significantly different from US-born Whites.

Table 3.

Predicted probabilities of food insecurity by immigration status-ethnicity, California Health Interview Survey 2020, Adult Respondents (N = 19,514).

	(1)		(2)		(3)
	Predicted prob.	SE	Predicted prob.	SE	Predicted prob.	SE
US-born Whites	0.056***	(0.01)	0.081***	(0.01)	0.082***	(0.01)
US-born Latinxs	0.127***	(0.01)	0.118***	(0.01)	0.119***	(0.01)
US-born Asians	0.041***	(0.01)	0.065***	(0.01)	0.063***	(0.02)
Naturalized Latinxs	0.143***	(0.02)	0.111***	(0.01)	0.114***	(0.01)
Naturalized Asians	0.057***	(0.01)	0.071***	(0.01)	0.071***	(0.01)
Non-citizen Latinxs	0.213***	(0.03)	0.126***	(0.02)	0.120***	(0.02)
Non-citizen Asians	0.140***	(0.03)	0.125***	(0.02)	0.125***	(0.02)

*** p < 0.001, ** p < 0.01, * p < 0.05, + p < 0.1. Note: Standard errors in parentheses. Model 1 adjusts for demographic variables, Model 2 adjusts for demographic and pre-pandemic SEP variables, and Model 3 adjusts for demographic, pre-pandemic SEP, and COVID-19 economic stressors.

Fig. 1.

Marginal predicted probability of food insecurity by immigration status-ethnicity. California Health Interview Survey 2020 Adult Respondents. N = 19,514 (based on Table 2, Model 3).

Table 4 presents the decomposition analysis of pre-pandemic SEP and COVID-19 economic stressors on the association between immigration status-ethnicity and food insecurity. Panel A shows the decomposition results of pre-pandemic SEP, and Panel B shows the results for COVID-19 economic stressors. Starting with Panel A, Latinxs have higher food insecurity than US-born Whites, regardless of immigration status. SEP remains a significant explanatory variable of the association between immigration status-ethnicity with food insecurity for all Latinx groups. However, the results for the direct effects reveal that immigration status and ethnicity categories also have an association independent of SEP. The percentage of mediation corroborates this finding as SEP accounts for a minimum of 44% to a maximum of 67% of the overall association between immigration status and ethnicity with food insecurity among Latinxs.

Table 4.

Decomposition of pre-pandemic socioeconomic position and COVID-19 economic stressors on immigration status-ethnicity categories and food insecurity, California Health Interview Survey 2020 adult respondents, N = 19,514.

Decomposition of effects	β (SE)	p-value
Panel A: Decomposition of pre-pandemic socioeconomic position on immigration status-ethnicity categories
Total effect of immigration status-ethnicity on food insecurity (ref. US-born White)
US-born Latinx	0.8965 (0.1377)	< 0.001
US-born Asian	−0.4837 (0.3110)	0.120
Naturalized Latinx	1.1721 (0.1914)	< 0.001
Naturalized Asian	−0.1184 (0.2370)	0.614
Non-citizen Latinx	1.5743 (0.2326)	< 0.001
Non-citizen Asian	0.8791 (0.2594)	0.001
Direct (unmediated) effect of immigration status-ethnicity on food insecurity (ref. US-born White)
US-born Latinx	0.5062 (0.1382)	< 0.001
US-born Asian	−0.3284 (0.3111)	0.291
Naturalized Latinx	0.4440 (0.1989)	0.026
Naturalized Asian	−0.1854 (0.2302)	0.440
Non-citizen Latinx	0.5212 (0.2479)	0.036
Non-citizen Asian	0.5729 (0.2634)	0.030
Indirect (mediated) effect of immigration status-ethnicity on food insecurity through pre-pandemic socioeconomic position (ref. US-born White)
US-born Latinx	0.3903 (0.1448)	0.007
US-born Asian	−0.1553 (0.1443)	0.282
Naturalized Latinx	0.7281 (0.1587)	< 0.001
Naturalized Asian	0.0660 (0.1442)	0.647
Non-citizen Latinx	1.0431 (0.1711)	< 0.001
Non-citizen Asian	0.3062 (0.1460)	0.036
Summary of mediation	%
Percent of Total effect due to pre-pandemic socioeconomic position (ref. US-born White)
US-born Latinx	43.53
US-born Asian	32.10
Naturalized Latinx	62.12
Naturalized Asian	−55.24
Non-citizen Latinx	66.89
Non-citizen Asian	34.83
Panel B: Decomposition of COVID-19 economic stressors on immigration status-ethnicity categories
Total effect of immigration status-ethnicity on food insecurity (ref. US-born White)
US-born Latinx	0.5105 (0.1381)	< 0.001
US-born Asian	−0.3308 (0.3111)	0.288
Naturalized Latinx	0.4370 (0.1990)	0.028
Naturalized Asian	−0.1795 (0.2401)	0.455
Non-citizen Latinx	0.5212 (0.2479)	0.036
Non-citizen Asian	0.5768 (0.2634)	0.029
Direct (unmediated) effect of immigration status-ethnicity on food insecurity (ref. US-born White)
US-born Latinx	0.5062 (0.1382)	< 0.001
US-born Asian	−0.3284 (0.3111)	0.291
Naturalized Latinx	0.4440 (0.1989)	0.026
Naturalized Asian	−0.1854 (0.2402)	0.440
Non-citizen Latinx	0.5212 (0.2479)	0.036
Non-citizen Asian	0.5729 (0.2633)	0.030
Indirect (mediated) effect of immigration status-ethnicity on food insecurity through COVID-19 stressors (ref. US-born White)
US-born Latinx	0.0043 (0.0374)	0.909
US-born Asian	−0.0024 (0.0374)	0.949
Naturalized Latinx	−0.0070 (0.0374)	0.851
Naturalized Asian	0.0060 (0.0373)	0.873
Non-citizen Latinx	0.0680 (0.0404)	0.092
Non-citizen Asian	0.0040 (0.0374)	0.916
Summary of mediation	%
Percent of Total effect due to COVID-19 economic stressors (ref. US-born White)
US-born Latinx	0.84
US-born Asian	0.72
Naturalized Latinx	−1.61
Naturalized Asian	−3.32
Non-citizen Latinx	11.54
Non-citizen Asian	0.69

Note: Panel A calculated effects account for demographic, familial, and COVID-19 stressors. Panel B calculated effects account for demographic, familial, and pre-pandemic socioeconomic position factors.

For Asians, we see a significant direct and indirect effect through SEP for non-citizen Asians relative to US-born Whites. There were no significant direct or indirect effects for US-born and naturalized Asians. Finally, SEP accounted for about 35% of the association between immigration status-ethnicity and food insecurity for non-citizen Asians.

Panel B shows the results for the decomposition of COVID-19 economic stressors. Overall, no significant decomposition effects of COVID-19 economic stressors existed for any immigration status and ethnic group. However, there was a marginally significant indirect effect (11.5%, p < 0.09) for non-citizen Latinxs only.

3.1. Sensitivity analysis

Appendix Table A displays the results of current employment as a covariate in the multivariable binary logistic regression of immigration status and ethnicity with food insecurity, including sociodemographic, familial, COVID-19 stressors, and pre-pandemic SEP. Overall, the magnitude and significance of coefficients and resulting conclusions are consistent with the main analyses.

4. Discussion

The COVID-19 pandemic revealed pre-existing disparities in food insecurity in which immigrants from minoritized racial-ethnic groups are disproportionately affected. Consistent with our hypothesis, we found significant heterogeneity in food insecurity by immigration status-ethnicity during COVID-19 and concluded with three key findings. First, non-citizen Asians and Latinxs of all immigrant categories were more food insecure than their US-born ethnic counterparts and US-born Whites. Second, pre-pandemic SEP explained most of the association between immigration status-ethnicity and food insecurity relative to COVID-19 economic stressors. Only among non-citizen Latinxs did COVID-19 economic stressors marginally explain food insecurity. Third, regardless of immigration status, Latinxs were more food insecure than their Asian counterparts, except for non-citizen Asians. Our results align with previous work on the intersection of immigration status/ethnicity and inequities in food insecurity in California (Walsemann et al., 2017).

Based on our multivariate analysis, we found a critical divide between US-born individuals and their non-citizen counterparts. Among Asians, non-citizens fared worse than US-born and naturalized Asians. For Latinxs, non-citizens were also more likely to be food insecure than their US-born and naturalized counterparts. Notably, when we control for COVID-19 economic stressors (in Model 3), the disparities in food insecurity between Latinx non-citizens and US-born Latinxs attenuate significantly, suggesting that the impact of COVID-19 economic stressors is greater for non-citizen Latinxs. Non-citizen Latinxs may be more vulnerable to the economic effects of the pandemic as this group makes up the largest share of the state's undocumented population (Hayes and Hill, 2017). Immigrants were more likely to have lost their jobs during the pandemic and remained ineligible for unemployment benefits (Kochhar and Bennett, 2021). Undocumented immigrants' vulnerability is compounded by their exclusion from federal social programs, including SNAP, unemployment, and Economic Stimulus Payments, which aim to help Americans weather the economic effects of the pandemic.

Further, an anti-immigrant socio-political environment may deter immigrants who qualify for relief from accessing these aids due to fear of becoming a public charge. For example, regardless of the child's or parent's immigration status, children from low-income families are eligible for school meals and associated interventions such as Pandemic-EBT; low-income women and their young children also qualify for Special Supplemental Nutrition Program for Women, Infants, and Children status regardless of immigration status. It may be possible that anti-immigrant sentiment prevents mixed-status households from accessing the aforementioned benefits, resulting in de facto avoidance of the available nutrition programs. While the Public Charge rule was reversed in March 2021, misinformation and lack of clear information regarding the rule's validity may also cause confusion among immigrant communities. Ultimately, the covert exclusion of immigrants from COVID-19 economic and social aid translates into economic precarity, which increases the risk of becoming food insecure.

Consistent with Walsemann et al. (2017), disparities in food insecurity by immigration status-ethnicity attenuated when we adjusted for SEP. Pre-pandemic SEP accounted for most of the variation in food insecurity, especially among non-citizen Asians and Latinxs of all immigration statuses. For example, among non-citizen Latinxs and non-citizen Asians, SEP explained 66.9% and  34.8% of the variance in food insecurity, respectively. Conversely, COVID-19 economic stressors, such as job loss or reduced wages/work hours, explained 11.5% of the variation in food insecurity among non-citizen Latinxs. While SEP indicators are dominant predictors of food insecurity, our findings suggest that the economic effects of the pandemic also contribute to non-citizen Latinxs' higher likelihood of being food insecure. Latinxs comprise a large share of California's undocumented population relative to other ethnic groups. Along with pre-pandemic financial precarity, job loss and ineligibility for social safety net programs among undocumented Californians during the ongoing pandemic may have exacerbated food insecurity risks for non-citizen Latinxs. Overall, compared to the effects of COVID-19 economic stressors, pre-pandemic SEP is a stronger predictor of food insecurity during the ongoing COVID-19 economic downturn. It is also important to note that immigrant-ethnic disparities in food insecurity remained net of SEP, COVID-19 economic stressors, and key demographic covariates, suggesting other factors we do not account for are likely driving food insecurity. For instance, previous work suggests that racism may play a role in generating disparities in food insecurity across racial-ethnic groups due to systemic economic inequality, structural racism, and personally-mediated discrimination (Odoms-Young and Bruce, 2018; Morales et al., 2021).

Lastly, naturalized and US-born Asians had similar levels of food insecurity as US-born Whites. We did not find any differences in food insecurity between US-born, naturalized Asians, and US-born Whites in our models. Non-citizen Asians had a similar probability of being food insecure as non-citizen Latinxs, suggesting that immigration status is a fundamental determinant of food insecurity above and beyond ethnicity. During the COVID-19 pandemic, Asian and Latinx non-citizens are more likely to be food insecure, suggesting that the broader anti-immigrant climate, public charge fears, and immigration-based eligibility restrictions for COVID-19 relief may be driving food insecurity. Further, naturalization does not confer the same protection against food insecurity among Latinxs as it does for Asians. Regardless of immigration status, Latinxs were more likely to be food insecure. Structural racism, in the form of differential access to opportunities and risks, may be driving this relationship for Latinxs in California (Odoms-Young and Bruce, 2018).

5. Limitations

Our findings should be considered in light of some limitations. First, the non-citizen category could not be disaggregated due to the public nature of the 2020 CHIS dataset. The non-citizen category is heterogeneous and includes undocumented immigrants, lawful permanent residents, and non-immigrant visa holders (e.g., those on temporary work or student visas). If we could further disaggregate this non-citizen category, we would predict that food insecurity would be even greater among those with more precarious immigration statuses like undocumented immigrants and temporary work visa holders. Indeed, previous studies using CHIS have shown gradients of health advantages among the non-citizen category (Bacong and Sohn, 2021). Despite our inability to disaggregate the non-citizen category, our study provides a rapid assessment of food insecurity by immigration status and ethnicity during the COVID-19 pandemic and invites future studies among these groups.

Second, the cross-sectional design limits our ability to make causal inferences. Although mediation analyses indicate that pre-pandemic SEP factors account for more of the association between immigration status-ethnicity and food insecurity than COVID-19 economic stressors, we cannot establish causal relationships. Third, California is not representative of the entire US, as the state has relatively more immigrant-inclusive social policies than other states (De Trinidad Young et al., 2018). Thus, our results may be a conservative estimate of the extent to which immigrants and racial/ethnic groups experience food insecurity.

Future studies could use imputation methods to more precisely disaggregate immigrant subgroups to examine the relationship between immigration status/race and food insecurity (i.e., undocumented, LPR). Given the heterogeneity in Asian and Latinx ethnic groups, it remains crucial to examine the possible differing experiences of Southeast Asians, Pacific Islanders, Mexican, Central Americans, and Puerto Ricans. Lastly, the omission of citizen and non-citizen non-Latinx Black Californians limits the generalizability of our findings to other states where the non-Latinx population makes up a large proportion of impoverished residents. Future work based in areas with a larger share of non-Latinx Black individuals could include this racial-ethnic group in the study sample.

6. Conclusions

We found critical within-group differences in food insecurity among Asians and Latinxs by immigration status. Non-citizen Asians and Latinxs of all immigration statuses were more likely to be food insecure than US-born Whites. Our study also provides novel information regarding drivers of food insecurity during COVID-19 among a representative sample of Asian and Latinxs immigrants in California. We decompose the variation in food insecurity and compare the role of pre-pandemic SEP versus novel COVID-19 economic factors. Our key findings include the obstinate role of pre-pandemic SEP in food insecurity for all Latinxs and non-citizen Asians and the additional burden of COVID-19 job loss and decreased wages for non-citizen Latinxs.

An upstream approach to address disparities in food insecurity among non-citizens includes a path to citizenship, which indirectly helps gain access to quality employment opportunities and social safety nets when needed. To prevent the widening disparities in food insecurity, local, state, and federal levels of government have the legal power to include non-citizens in the social safety net, as was done before the 1970s when immigration status was not an eligibility criterion for public assistance (Fox, 2016). Overall, the groups most affected by the COVID-19 pandemic will continue to experience the worst outcomes during future social shocks; thus, it is essential to bolster the social safety net to support non-citizens and marginalized groups.

CRediT authorship contribution statement

Alein Y. Haro-Ramos: Conceptualization, Methodology, Software, Formal analysis, Writing – original draft, Writing – review & editing, Funding acquisition. Adrian M. Bacong: Writing – review & editing, Methodology, Software, Formal analysis.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary material

Data availability

Data will be made available on request.