OBJECTIVES: Both obesity and food insecurity are important public health problems facing young children in the United States. A lack of affordable, healthy foods is one of the neighborhood factors presumed to underlie both food insecurity and obesity among children. We examine associations between local food prices and children’s BMI, weight, and food security outcomes.
METHODS: We linked data from the Early Childhood Longitudinal Study-Birth Cohort, a nationally representative study of children from infancy to age 5, to local food price data from the Council for Community and Economic Research (C2ER) Cost-of-Living Index (n = 11 700 observations). Using ordinary least squares (OLS), linear probability, and within-child fixed effects (FE) models, we exploit the variability in food price data over time and among children who move residences focusing on a subsample of households under 300% of the Federal Poverty Level.
RESULTS: Results from ordinary least squares and FE models indicate that higher-priced fruits and vegetables are associated with higher child BMI, and this relationship is driven by the prices of fresh (versus frozen or canned) fruits and vegetables. In the FE models, higher-priced soft drinks are associated with a lower likelihood of being overweight, and surprisingly, higher fast food prices are associated with a greater likelihood of being overweight.
CONCLUSIONS: Policies that reduce the costs of fresh fruits and vegetables may be effective in promoting healthy weight outcomes among young children.
- C2ER —
- Council for Community and Economic Research
- CBSA —
- Core-Based Statistical Area
- CI —
- confidence interval
- COLI —
- Cost of Living Index
- ECLS-B —
- Early Childhood Longitudinal Study-Birth Cohort
- FE —
- fixed effects
- FPL —
- Federal Poverty Level
- LPM —
- linear probability model
- OLS —
- ordinary least squares
What’s Known on This Subject:
A growing body of research suggests that the food environment affects children’s weight. Specifically, living in areas with higher-priced fast foods and soda is associated with lower weight and BMI, whereas higher fruit and vegetable prices demonstrate the opposite association.
What This Study Adds:
Using longitudinal data on lower-income young children, this study finds that higher-priced fruits and vegetables are associated with higher child BMI, but not food insecurity, and that this relationship is driven by the prices of fresh fruits and vegetables.
Both under- and overnutrition are important public health problems facing young children in the United States. In 2011, ∼20.6% of households with children in the United States were food insecure, defined by the US Department of Agriculture as having “limited or uncertain availability of nutritionally adequate and safe foods or limited or uncertain ability to acquire acceptable foods in socially acceptable ways.”1,2 More than 26% of 2- to 5-year-old children were considered overweight (defined as having a BMI above the 85th percentile by age and gender) in 2009–2010, up from 21% in 1999–2000.3 Being food insecure and/or overweight during early childhood has negative effects on children’s short- and long-term health, social, and economic outcomes.4–9
A lack of affordable, healthy foods is one of the neighborhood factors presumed to underlie both food insecurity and obesity among children.10,11 Although general food prices (ie, price per calorie) trended downward in recent decades, particularly the prices of snacks and sugar-sweetened beverages, the real prices of restaurant meals and fruits and vegetables increased,12 with fruit and vegetable prices increasing by 17% between 1997 and 2003 alone.13 Experimental work has found that children decrease their consumption of certain foods when the price is increased.14 Living in areas with higher-priced fast foods and soda is associated with lower body weight and BMI, whereas higher fruit and vegetable prices demonstrate the opposite association.15–22 These relationships appear to be larger among low-income children as compared with their higher-income counterparts,15,17,18 presumably because their families have less disposable income. With a tight budget constraint, a family may purchase more poorer-quality, energy-dense foods,23–25 which cost less per calorie than more nutritious foods,26,27 although not by weight or average portion.28
Despite the importance of adequate nutrition during early childhood, little research has examined how food prices relate to weight and food insecurity outcomes during early childhood,29 and, with few exceptions,16,17,30 most studies have estimated cross-sectional associations between food prices and child outcomes. Further, previous research has not isolated fresh fruits and vegetables, whose prices vary more than frozen or canned options, and little research has examined soft drink prices, important considering that sugar-sweetened beverages account for nearly 15% of children’s daily caloric intake31 and soft drinks can have negative effects on children’s health.32
This study estimates associations between local food prices and the weight and food insecurity outcomes of children from infancy to 5 years of age. We hypothesize that (1) high-priced fruits and vegetables and low-priced fast food and soft drinks may contribute to a greater likelihood of being overweight and a higher BMI; (2) high prices for fruits and vegetables, fast food, and soft drinks may contribute to a greater likelihood of being food insecure; and (3) the prices of fresh fruits and vegetables will be more strongly associated with outcomes than frozen or canned fruits and vegetables.
We use child-level data from the Early Childhood Longitudinal Study-Birth Cohort (ECLS-B) linked with city-level data from the Council for Community and Economic Research (C2ER), formerly known as ACCRA Cost of Living Index. The ECLS-B is a longitudinal dataset of approximately 10 700 children and is nationally representative of children born in the United States in 2001 (Sample sizes are rounded to the nearest 50 according to data use requirements.). The ECLS-B collected data when children were 9 months of age (2001–2002), 2 years of age (2003–2004), 4 years of age (preschool: 2005–2006), and at 2 waves of kindergarten entry (2006–2008). This study uses the first 4 waves of data, excluding the second kindergarten entry wave (2007–2008) because wave 5 only included a smaller sample of children who repeated kindergarten or were just entering kindergarten. At each wave, information about child and family characteristics and residential zip codes were collected through interviews with parents and child assessments.
The C2ER Cost of Living Index (COLI) data are collected quarterly by the Council for Economic Research (http://www.coli.org) from more than 300 Core-Based Statistical Areas (CBSA; previously Metropolitan Statistical Areas). CBSAs constitute relatively large geographic areas, and exclude certain areas of the country, particularly rural areas. Despite its limitations, the C2ER food price data have been used in >15 studies since 2002 and correlate highly with other sources of food price data.12 For this study, data from 2001 through 2007 were merged with the ECLS-B. Approximately 5700 children (54%) lived in CBSAs that were matched to C2ER food price data for at least 1 wave, comparable to previous research using these data.18
The unit of analysis in the study is the child in a given wave. We limit our sample to observations of children with nonmissing data on food price and control variables who reside in households with income below 300% of the Federal Poverty Level (FPL; $61 950 for a family of 4 in 2007) at any wave, as the outcomes of children living in lower-income families are more likely to be affected by food prices. The sample size varies by the dependent variable and ranges from 6450 to 11 700. Our analysis sample averaged lower BMI z scores and rates of overweight than those low-income children excluded because of missing C2ER data, not surprising given higher rates of child obesity in rural areas.33
BMI z scores, overweight status, and food insecurity served as the dependent variables. At the 2-year, preschool, and kindergarten entry waves of data collection, anthropometric measures of children’s weight and height were collected twice in a laboratory setting by trained data collectors. If the 2 measures were >5% apart, a third measurement was taken. The 2 closest measurements were averaged. By using the Centers for Disease Control and Prevention standards,34 standardized BMI z scores were generated to allow for comparisons across age and gender. The binary outcome of overweight (having a BMI at or above the 85th percentile for age and gender) is used (1 = overweight). We exclude the weight outcomes of children <24 months.
At all data waves, participating households were asked about their experiences of food insecurity over the past 12 months (since the previous interview at the 2-year wave) using the 18-question Core Food Security Module created by the US Department of Agriculture. Their responses were used to generate 2 binary indicators of household food insecurity: one that includes both low and very low food security (1 = low or very low food security), and another indicating that the household has very low food security (1 = very low food security).
Our independent variables include the average prices of the following items measured in the C2ER data: (1) 6 fruits and vegetables (potatoes, bananas, lettuce, canned sweet peas, canned peaches, and frozen corn), (2) 3 fast foods (the average price of a McDonald’s quarter-pounder with cheese; the average price of an 11-inch to 12-inch thin-crust regular cheese pizza at Pizza Hut and/or Pizza Inn; and the average price of a fried chicken drumstick and thigh at Kentucky Fried Chicken and/or Church’s Fried Chicken), and (3) a soft drink (2-L bottle of Coca-Cola). For some analyses, we separated the fruits and vegetables into (1) fresh fruits and vegetables (potatoes, bananas, and lettuce) and (2) frozen and canned fruits and vegetables (canned peaches, canned sweet peas, and frozen corn). Prices are collected from food stores and restaurants that serve households across the income distribution. Because food prices are not collected for every CBSA at each quarter, measures of average annual food price indices were calculated for each category. Each of our price measures was divided by the annual C2ER overall cost-of-living composite index and inflation-adjusted to 2008 dollars. We then standardized these ratios relative to entire C2ER sample, generating z scores (x¯ = 0, SD = 1). Average annual food prices and the standardized ratios of the national C2ER sample are provided in Table 1, which illustrates considerable price variability.
We control for potential confounding factors that may be associated with both local food prices and outcomes. Respondents (96.5% of whom were the mothers of the sampled children) reported the child’s gender, race/ethnicity (non-Hispanic white, non-Hispanic black, Hispanic, other), whether the child was a multiple birth, age in months, and birth weight (in kg), and mothers’ prepregnancy weight (in kg). We use binary indicators of respondent-reported parental education at wave 1 (neither parent graduated high school, at least 1 parent graduated high school, and at least 1 parent graduated college). At each wave, respondent-reported family income and the number of children and adults living in the household were used to code the household’s income-to-needs ratio based on the FPL. Also at each wave, respondents reported whether anyone in their household owned a car and their own employment status and weekly work hours (mother worked 35 or more hours per week, mother worked fewer than 35 hours per week, mother was not employed). The household’s geographic region (Northeast, Midwest, West, South), urbanicity (urban or rural/suburban), and data wave were also controlled.
A series of ordinary least squares (OLS) and linear probability models (LPM) were estimated. To address the research aims, OLS regression was used to estimate how local food prices influence children’s BMI z scores, controlling for child, maternal, and household characteristics. For the binary dependent variables of overweight and food insecurity, analogous LPMs were estimated. Logit models were also estimated. Results were not substantially different (available on request). SEs were clustered at the CBSA level.
Although many covariates are included, it is likely that the food price estimates generated are biased. To further limit bias, within-child fixed effects (FE) models were also estimated, exploiting the variation in prices over time for each child and among the children in the sample (13%) who moved across CBSAs between the 9-month and kindergarten entry waves.
We re-ran these OLS, LPM, and FE models, separating the fruits and vegetables price index into separate indices for fresh fruits and vegetables and frozen or canned fruits and vegetables. Measures of effect size (d) for BMI z scores as a continuous dependent variable are presented (coefficient/SD of the BMI z score [1.34]).
The analysis sample’s descriptive statistics, pooled across waves, are displayed in Table 2. On average, children’s BMIs were about 0.5 SD above Centers for Disease Control and Prevention recommendations. About 31% of children were overweight, and about 15% of children lived in households that reported low or very low food security. Overweight children faced higher average annual fruit and vegetable, fast food, and soft drink prices than their nonoverweight peers, but the standardized price ratios did not differ. Food-insecure households faced lower average fruit and vegetable prices than food-secure households.
Table 3 displays the primary model results for the OLS and FE models. Consistent with hypotheses, standardized fruit and vegetable price ratios are associated with higher child BMI z scores. In the OLS models, a 1-unit increase (1 SD) in the standardized ratio of average annual fruit and vegetable prices, or a $0.24 increase, is associated with an increase of 0.088 (95% confidence interval [CI] 0.028 to 0.148) in a child’s BMI z score (d = 0.066). The magnitude of this association is about two-thirds that of the association between living below the poverty line and BMI. Fast food and soft drink prices are unrelated to children’s BMI z scores or rates of overweight or household food insecurity in the LPM models.
Results from the within-child FE models indicate that a $0.24 (1 SD) increase in the price of fruits and vegetables is associated with a 0.107 (95% CI 0.026 to 0.188) increase in children’s BMI z scores (d = 0.080), and a 0.8 (95% CI 0.001 to 0.015) percentage point increase in the likelihood that a household experiences very low food security. In addition, a $0.17 increase in the price of soft drinks is associated with a 2.5 (95% CI –0.049 to –0.001) percentage point decrease in the likelihood a child is overweight. Counter to expectations, a $0.40 increase in the price of fast foods is associated with a 5.9 (95% CI 0.026 to 0.092) percentage point increase in the likelihood that a child is overweight in the FE models only.
Table 4 displays models with separate price indices for fresh versus frozen and canned fruits and vegetables. Results suggest that the prices of fresh fruits and vegetables are driving the association between fruits and vegetables and children’s BMI z scores (d = 0.092 in the OLS models, d = 0.117 in the FE models) and very low food security (in the FE models only). A $0.38 increase in the price of fresh fruits and vegetables is also associated with a 2.5 (95% CI 0.000 to 0.051) percentage point increase in the likelihood that a child is overweight in the LPM results. The same increase in the price of fresh fruits and vegetables is associated with a 0.8 (95% CI 0.000 to 0.015) percentage point increase in the likelihood that a household experiences very low food security in the FE results.
We estimated 4 sensitivity analyses (results available in Supplemental Information Tables 5–8) that did not change our results. First, state-level sales tax rates and whether food was exempt from sales tax, gathered from the Bridging the Gap program, were added as controls. Second, we modified our independent variable from the ratio of prices to the overall cost of living to the raw food prices (inflated to 2008 dollars), controlling for the overall cost-of-living ratio. Third, we added a parent-reported measure of children’s amount of screen time (watching television or DVDs). Finally, we tested lagged associations between our food price measures and child outcomes.
The goal of this study was to estimate associations between local food prices and young children’s weight and food insecurity outcomes. Consistent with previous research,15,20,35 children living in areas with higher-priced fruits and vegetables averaged higher measures of standardized BMI scores, compared with their peers in areas with lower-priced fruits and vegetables. We found that these associations are driven by changes in the prices of fresh fruits and vegetables rather than frozen or canned. The magnitude of this association is considerable; a $0.38 increase in the average annual price of fresh fruits and vegetables is linked with about a one-ninth to one-eleventh of an SD increase in children’s BMI z scores in our OLS and FE models. Although these changes reflect relatively small increases in children’s BMI measures, the corresponding price changes are relatively small as well, and the range of food prices across CBSAs suggests that residential moves may expose children to areas with substantial variation in prices. Unfortunately, our subsample of children who moved to different CBSAs and have full price data were limited (fewer than 300); future research should exploit residential moves as a means for testing the relationship between food prices and child weight outcomes.
Consistent with hypotheses, higher soft drink prices were associated with a decrease in the likelihood of being overweight in the FE models. Contrary to expectations, higher fast food prices were associated with an increase in the likelihood of being overweight in the FE models. The FE models are more limited in their sample size, and the inclusion of a selective subsample may underlie the differences between the OLS and LPM models and the FE models. Alternatively, this may be a result of endogeneity; fast food outlets may respond to increased demand or preferences for fast food with higher prices. Indeed, previous research indicates that fast food locales have substantial independent control over their prices.36 Further, although the literature on the relationship between fruit and vegetable prices and child BMI is relatively consistent, the research on fast food prices and child weight outcomes is mixed.15,20,37
Also in contrast with expectations, food prices were largely unrelated to household-level food insecurity, with 1 exception: an increase in the average price of fresh fruits and vegetables was associated with a small increase (0.8 percentage points) in the likelihood of experiencing very low food security in the FE models only, suggesting that even small increases in the price of healthy foods may increase food insecurity among those already at risk. More research using different categories of foods or neighborhood-based measures of prices is needed.
This study has several limitations. First, our analyses cannot reveal causal associations between food prices and health outcomes. It is likely that families select into cities or neighborhoods that match their food and cost-of-living preferences. Although our FE models limit bias from unobserved, stable characteristics, they cannot address bias from time-varying characteristics or from reverse causality. Second, our measure of food security is respondent-reported and was asked about the previous 12 months (or the previous data collection wave), which may not align with our calendar year measures of food prices. Third, although the ECLS-B is a nationally representative cohort, the particular sample we use for our analysis is not nationally representative. Fourth, our food price measures are assessed at relatively large geographic areas, focus on urban areas, and are limited in the food items assessed. Finally, our food price measures represent annual averages, masking seasonal variability.
Despite these limitations, this study identifies significant associations between food prices and outcomes and sheds light on promising policies. Health professionals working with low-income children in areas with high-cost foods should be aware of the potential for heightened risk of overweight or obesity, and may consider referring households to programs that provide reduced-price healthy foods. Results suggest that policies that subsidize the cost of fresh fruits and vegetables may be effective in improving the health and weight outcomes of young children. The Supplemental Nutrition Assistance Program (formerly known as the Food Stamp Program) is testing new initiatives, including financial incentives that reduce the costs of fruits and vegetables for recipients. Massachusetts’ Healthy Incentives Pilot and New York City’s financial incentives for fruits and vegetables at farmers’ markets are 2 examples. More research on the effects of these new initiatives and how public nutrition assistance buffers or mitigates the effects of food prices on children’s outcomes is needed.
The authors are grateful for helpful comments from participants at colloquia presented at the Population Association of America Annual Conference, the Association for Public Policy Analysis and Management Annual Conference, the National Association for Welfare Research and Statistics Annual Conference, the USDA Economic Research Service Grantee Conference, the University of Chicago, and the University of Wisconsin-Madison.
- Accepted December 11, 2013.
- Address correspondence to Taryn W. Morrissey, PhD, School of Public Affairs, American University, 4400 Massachusetts Ave, NW, Washington, DC 20016. E-mail:
Dr Morrissey led the conceptual design of the study, secured the project funding as the principal investigator, and drafted most of the initial manuscript; Dr Jacknowitz contributed to the project’s conception, secured the project funding as the co-principal investigator, and wrote and edited sections of the manuscript; Ms Vinopal carried out the initial and revised analyses, and reviewed and revised the manuscript; and all authors approved the final manuscript as submitted.
The views and opinions expressed herein are those of the authors and do not necessarily reflect those of the Institute for Research on Poverty or the Food Assistance and Nutrition Research Program, the Economic Research Service, or the US Department of Agriculture.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: This research was funded through the Institute for Research on Poverty RIDGE Center for National Food and Nutrition Assistance Research, which is supported by the Food Assistance and Nutrition Research Program of the US Department of Agriculture’s Economic Research Service.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
- ↵Coleman-Jensen A. Household Food Security in the United States in 2011. Washington, DC: USDA; 2012
- ↵Bickel G, Nord M, Price C, Hamilton W, Cook J. Guide to Measuring Household Food Security Revised 2000. Alexandria, VA: USDA, FNS; 2000
- Sturm R
- Crosnoe R,
- Muller C
- Preventing Childhood Obesity IOM
- Rose D
- ↵Christian T, Rashad I. Trends in US food prices, 1950–2007. Econ Hum Biol. 2009;7(1):113–120. Available at: www.sciencedirect.com/science/article/pii/S1570677X08000634. Accessed June 7, 2012
- Cawley J
- Epstein LH,
- Handley EA,
- Dearing KK,
- et al
- ↵Powell L, Chaloupka FJ. Economic contextual factors and child body mass index. NBER volume on Economic Aspects of Obesity. Cambridge, MA: National Bureau of Economic Research; 2010. Available at: www.nber.org/chapters/c11818.pdf. Accessed January 3, 2014
- ↵Sturm R, Datar A. Body mass index in elementary school children, metropolitan area food prices and food outlet density. Public Health. 2005;119(12):1059–1068. Available at: www.ncbi.nlm.nih.gov/pubmed/16140349. Accessed October 7, 2012
- Sturm R,
- Powell LM,
- Chriqui JF,
- Chaloupka FJ
- ↵Han E, Powell LM. Effect of food prices on the prevalence of obesity among young adults. Public Health. 2011;125(3):129–135. Available at: www.ncbi.nlm.nih.gov/pubmed/21272902. Accessed October 7, 2012
- Drewnowski A,
- Specter SE
- ↵Monsivais P, Mclain J, Drewnowski A. The rising disparity in the price of healthful foods: 2004–2008. Food Policy. 2010;35(6):514–520. Available at: http://linkinghub.elsevier.com/retrieve/pii/S030691921000076X. Accessed October 7, 2012
- ↵Carlson A, Frazão E. Are Healthy Foods Really More Expensive? It Depends on How You Measure the Price. Washington, DC: US Department of Agriculture, Economic Research Service; 2012. Available at: www.ers.usda.gov/publications/eib-economic-information-bulletin/eib96.aspx. Accessed January 3, 2014
- ↵Gundersen C, Kreider B, Pepper J. The economics of food insecurity in the United States. Applied Economic Perspectives and Policy. 2011;33(3):281–303. Available at: http://aepp.oxfordjournals.org/lookup/doi/10.1093/aepp/ppr022. Accessed November 26, 2012
- Kimbro RT,
- Rigby E
- ↵Wang YC, Bleich SN, Gortmaker SL. Increasing caloric contribution from sugar-sweetened beverages and 100% fruit juices among US children and adolescents, 1988-2004. Pediatrics. 2008;121(6). Available at: www.pediatrics.org/cgi/content/full/121/6/e1604
- ↵Liu J, Bennet KJ, Harun N, Zheng X, Probst JC, Pate RR. Overweight and physical inactivity among rural children aged 10-17: a national and state portrait. Columbia, SC: South Carolina Rural Health Research Center; 2007. Available at: http://rhr.sph.sc.edu/report/%287-1%29Obesity%20ChartbookUpdated10.15.07-secured.pdf. Accessed January 3, 2014
- ↵Lafountaine F. Pricing Decisions in Franchised Chains: A Look at the Restaurant and Fast-Food Industry. Cambridge, MA: National Bureau of Economic Research; 1995
- ↵Powell LM. Fast food costs and adolescent body mass index: Evidence from panel data. J Health Econ. 2009;28(5):963–970. Available at: www.sciencedirect.com/science/article/pii/S0167629609000678. Accessed June 7, 2012
- Copyright © 2014 by the American Academy of Pediatrics