Did the New School Meal Standards Improve the Overall Quality of Children's Diets?

Introduction

Together, the School Breakfast (SBP) and National School Lunch Programs (NSLP) represent the second largest form of federal food assistance in the United States. These two programs operate in nearly 100,000 institutions, reaching approximately 30 million students during lunchtime every school day and 15 million students during breakfast. Schools receive federal cash reimbursements for each meal served, totaling $17 billion for the 2018–2019 school year, provided they meet certain nutritional standards.

The passage of the Healthy Hunger-Free Kids Act of 2010 (HHFKA; US Public Law 111-296 2010) represented the first major shift in nutritional standards in fifteen years. The HHFKA realigned school meal standards to the Dietary Guidelines for Americans, the federal government's official recommendations for healthy eating. Officially in effect for the 2012–2013 school year, the new standards require increased availability of fruits, vegetables, and whole grains while reducing sodium and disallowing whole milk. The new standards also introduced calorie minima and maxima by age groups. Nonreimbursable foods sold at school, known as competitive foods, must also meet increased nutritional standards beginning in 2014–2015. Further institutional details of the school meal programs, along with the specifics of the policy change, are discussed in the next section.

The objective of this study is to better understand how changes in nutritional standards for federally subsidized school meals have impacted the quality of children's diets. Several earlier studies conducted at the local level found the new school meal standards had positive average impacts on the quality of specific foods offered (Terry-McElrath, O'Malley, and Johnston 2015), selected (Amin et al. 2015; Johnson et al. 2016), and consumed (Cohen et al. 2014; Cullen, Chen, and Dave 2015). More recently, national-level results are emerging. Gearan and Fox (2020) found the quality of breakfast and lunch offerings, not necessarily foods consumed, to be higher in 2014–2015 as compared to those served in 2009–2010. Lin, Guthrie, and Smith (2019) showed that the consumption of whole grains from school-prepared meals increased postimplementation of the HHFKA. A recent report commissioned by the USDA (2019) found the national average quality of food consumed from school-prepared lunches to be higher than home-prepared lunches during the 2012–2013 school year; however, this study did not examine preimplementation comparisons due to data limitations.

No study has examined differential impacts of the HHFKA across the dietary spectrum—that is, beyond the average impact—by comparing program effects for children who typically eat poorly versus those who typically consume higher-quality diets. Program evaluation beyond the average impact can be particularly important when a heterogeneous population is exposed to a rather homogenous policy (Heckman, Smith, and Clements 1997; Heckman 2005). Allowing for this type of heterogeneity is particularly important given that the federal government sets minimal nutritional standards that apply to all schools, creating a relatively homogeneous standard for a rather diverse population.

Indeed, previous work using data predating the HHFKA has shown school food programs increase the quality of diets for children who are most vulnerable (Bhattacharya, Currie, and Haider 2006; Howard and Prakash 2012; Smith 2017), while having little-to-no effects (or even negative effects) for children who typically eat relatively well outside of school (e.g., higher-income children) (Gleason and Suitor 2003, Mancino, Todd, and Lin 2009, Campbell et al. 2011). We examine how the relationship between school food and the overall quality of children's diets has evolved over time, with a focus on the transition into the HHFKA era.

This study uses a nationally representative cohort of children from the National Health and Nutrition Examination Survey (NHANES). The data collect two days of twenty-four-hour dietary intakes. The first day is administered in person, and the second interview is three to ten days later in a follow-up telephone interview. Following previous studies (Bowman et al. 2004; Mancino, Todd, and Lin 2009; Powell and Nguyen 2013; Smith 2017), we leverage this short two-day panel using a fixed effects approach to control for unobserved factors associated with selection into school meal programs (e.g., the school environment and parental characteristics). From the detailed food recalls, we calculate dietary scores using the Healthy Eating Index (HEI).

We focus on pre- and postpolicy effects of consuming school food by comparing 2009–2012 to 2013–2016.¹ On average, we find the consumption of a school meal, representing roughly 33% of daily calorie consumption, in lieu of a home-prepared meal increased dietary quality by 3.21% (or 1.56 HEI points) during 2009–2012. After the implementation of the HHFKA, the average school-meal effect more than doubled to 7.74% (or 3.78 HEI points).

One of the stated purposes of the Act was “to increase access to healthy food for low-income children,” according to a fact sheet distributed by the White House in December 2010 (The White House, Office of the Press Secretary 2010). The income disparity with regards to the school-meal effect was marked in the pre-HHFKA period: no returns to a school meal for high-income students (−0.8%), but large and significant returns among their low-income counterparts (6.8%). This finding reflects previous literature (Bhattacharya, Currie, and Haider 2006; Howard and Prakash 2012; Smith 2017) showing that virtually all the dietary gains to school food in the pre-HHFKA period was concentrated among low-income students. Post-HHFKA, the income disparity significantly shrunk due to the large gains among higher-income students (4.4%), as compared to the gains observed among low-income students (10.6%). In other words, higher-income children benefited more from the HHFKA whereby the school-meal effect increased substantially from −0.8% to 4.4%. The low-income group's increase from 6.8% to 10.6%, while notable, is not a statistically significant increase.

Policymakers tend to place an increased focus on the most vulnerable subpopulations of children. Yet, not all children who eat poorly are low income. To this end, we estimate the impact of a school meal across the entire distribution of dietary quality using a fixed-effects quantile regression (Powel 2016). This allows us to focus on the lower tail of the dietary quality distribution. Preimplementation of the new school meal standards (2009–2012), the average impact was mainly driven by the lower half of the distribution, where program effects where slightly larger than the average affects (about 2 HEI points). However, within the upper portion of the preimplementation distribution we find no significant effects with point estimates falling below zero above the eightieth percentile. These results echo those found in Smith (2017), who examined the pre-HHFKA period 2005–2010.

Postimplementation of the HHFKA in the years 2013–2016, we see not only larger impacts in the lower tails as compared to pre-implementation, but also see these impacts persist into the upper tails. Specifically, when children substitute away from a home-prepared meal towards a school-prepared meal, the entire distribution of dietary quality shifts by about three to four HEI points. In fact, at all points in the distribution we reject the null of no effect. We also find that the quality of home-prepared meals remained relatively stable over 2009–2016, implying the change in the school-meal effect is operating off school food and not home food. In total, these findings imply that the HHFKA is a first-order policy improvement with regards to the quality of foods being consumed by school children.²

Institutional Background

The National School Lunch Program (NSLP) was enacted in 1946 on the premise of providing “an adequate supply of food” (US Public Law 79-396 1946); the School Breakfast Program (SBP), which included the additional emphasis on “adequate nutrition” (US Public Law 94-105 1975), was enacted in 1975. This shift from simply providing food to providing higher-quality food has persisted through the decades and continues today (Gunderson 1971; Ralston et al. 2008).

Schools receive federal reimbursements for each meal served provided they meet a certain set of nutritional standards. Moreover, reimbursements are based on household income. Children who live in households with income below 130% of the federal poverty guidelines can receive a free school meal, while those between 130% and 185% pay a (substantially) reduced price. All other children pay the full price. However, it must be noted that even the full-price meals are federally subsidized.³

In 1996, nutritional standards for school meals were linked the 1995 Dietary Guidelines for Americans (DGA). The DGA are the federal government's official recommendations for healthy eating. Since 1980, the DGAs have been updated every five years. However, subsequent iterations of the DGA in the early 2000s were not incorporated into the nutritional standards for schools to receive federal reimbursements. This changed with the Healthy Hunger Free Kids Act of 2010, which mandated the United States Department of Agriculture (USDA), through the Food and Nutrition Service (FNS), update school meal regulations to conform to the 2005 DGAs. The new standards required more fruit and vegetable offerings, as well as increased whole grain offerings, while limiting sodium and whole milk. The new standards also set calorie maxima and minima. The standard went into effect in the 2012–2013 school year for lunches and in 2013–2014 for breakfasts.

In December 2018, FNS published “Final Rule: Child Nutrition Program Flexibilities for Milk, Whole Grains, and Sodium Requirements.” The rule codified three menu-planning changes: allowing local operators to permanently offer flavored low-fat milk; requiring half (rather than all) of the weekly grains be whole-grain-rich; and providing more time for gradual sodium reduction by retaining Sodium Target 1 through the end of school year (SY) 2023–2024, continuing to Target 2 in SY 2024–2025, and eliminating the Final Target that would have gone into effect in SY 2022–2023. The final rule went into effect on February 11, 2019. Our analysis considers the time period before implementation of the FNS interim and final rules on “Child Nutrition Program Flexibilities for Milk, Whole Grains, and Sodium Requirements.”

Data

We use a nationally representative cohort of children from the National Health and Nutrition Examination Survey (NHANES). NHANES gathers information on twenty-four-hour dietary recalls on two nonconsecutive days using computer-assisted recall methods to reduce misreporting (Moshfegh et al. 2008). The first recall (Day 1) is collected in-person during the medical exam. The second recall (Day 2) is collected three to ten days later over the telephone. Both interviews are conducted by a trained dietary interviewer with the aid of three-dimensional aids (e.g., measuring cups). For children under the age of six, interviews are obtained via a proxy, typically a parent; for children aged six to eleven, proxies are present to assist in the interview; all other students twelve years or older self-report dietary intakes. In most cases, the second day is with respect to a different day of the week as compared to Day 1.

The sample consists of children who report attending kindergarten through twelfth grade during the school year who had complete dietary intakes for both days. Following the previous literature, we exclude children whose schools do not offer a lunch (Gleason and Suitor 2003; Schanzenbach 2009; Millimet, Tchernis, and Husain 2010; Gundersen, Kreider, and Pepper 2012; Smith 2017). The main comparison will be the effect of substituting home-prepared food for school-prepared food pre-HHKFA (2009-12) and post-HHFKA (2013-16), which have 3,818 and 3,523 children reporting two days of intake, respectively.

We also examine differences by grade (elementary school versus middle/high school) and income status. We define low-income as households who fall below 185% of the federal poverty guidelines, as done elsewhere (Gundersen, Kreider, and Pepper 2012). The choice of 185% is intended to capture both free and reduced-price students. Results are robust to using alternative thresholds.

Our main outcome is the overall quality of daily food consumption. In this manner, we consider all foods consumed from all sources on a given day, which allows for the notion that children may vary their at-home diets based on what they consumed at school.

Distributional Descriptive Statistics

Figure 1 gives a sense of how eating school-prepared food may impact dietary quality at different points in the distribution, both prior to the HHFKA (left column) and after its implementation (right column). The two top panels plot out HEI scores by their respective quantiles conditional on consuming any calories from school. The bottom two panels of figure 1 report the differences between the diet-quality distributions conditional on school food consumption within each period. For example, prior to the HHFKA, the bottom half of the dietary quality distribution increased by about four HEI points when children consumed school-prepared food; this difference dissipates as we move to higher quantiles. Post-HHFKA, a similar shape is observed, although larger in magnitude: The bottom half of the distribution now shifts by about six HEI points before trailing off at higher quantiles. Thus, at least descriptively, we see: (i) larger dietary gains from consuming school food within the bottom of the outcome distribution in both periods, and (ii) an increase in the positive correlation of consuming school food at all points in the dietary quality distribution by roughly 2 HEI points.

Clearly, the associations shown in the bottom panels of figure 1 within each policy period are confounded. First, there is a selection issue: Children who choose to consume school food may differ based on unobservable characteristics that also drive overall diet quality. Secondly, there exists an omitted variables problem: Comparing school food consumed during the week to home-prepared food consumed on weekdays and weekends will overestimate the effect of school food. Moreover, figure 1 is only suggestive, because each distribution is conditional on a single food source, and it cannot additionally disentangle the effects of consuming food away from home. Therefore, the next section introduces our methodological approach to help alleviate these concerns. In short, the quantile regression approach aims to identify the results shown in the bottom panels of figure 1.

Methods

Results

Distributional Effects

Figure 2 shows the effect of substituting a home-prepared meal for a school-prepared meal at various points in the dietary quality distribution for all students.⁸ Prior to the implementation of the HHFKA during 2009–2012, the positive impact of a school meal was concentrated below the median at roughly two HEI points (or about 3–6% of the at-home diet, see Appendix table C.1). Above the median, we see no significant effect of a school meal, indicating home-prepared meals were of similar quality. Post-HHFKA, however, significant gains were made across the entire distribution dietary quality for the general population of students, especially at higher quantiles. In total, we see a roughly four-point constant increase due to the consumption of a school meal in the post-HHFKA period. This four-point increase in HEI translates into a 12% increase at the tenth percentile of the at-home dietary quality distribution and a 4% increase at the ninetieth percentile, all statistically significant. ⁹

Figure 3 splits the sample by grade level. For elementary school children, we can see that the average impact prior to the HHFKA was again mainly driven by the lower portion of the distribution. Post-HHFKA, the school-meal effect for younger students increased substantially at all points across the at-home dietary quality spectrum (Appendix figure D.1). Older students in middle and high school experienced even more notable gains: In the pre-HHFKA era there was virtually no difference between home-prepared and school-prepared meals in terms of their dietary quality. In the post-HHFKA era, however, school meals were of higher quality at all points in the distribution (Appendix figure D.1).

Figure 4 shows results by income level, using a threshold of 185% of the poverty guidelines. Lower-income students in the prepolicy period benefited from school meals across the board. Similar effects are seen in the postpolicy period. The over change is minimal as compared to higher-income students (Appendix figure D.1).

Policy Discussion

The above results should inform on-going policy discussions pertaining to reforming school meal standards. Historically, Congress has reauthorized child nutrition programs every five years, at which time program parameters are subject to revisions. In 2016, separate House and Senate bills were sent to committee (House Bill H.R.5003 2016; Senate Bill S.3136 2016), but neither made it to their respective full chambers for a vote. As such, no reauthorization has occurred since the passage of the 2010 HHFKA.

Reauthorization is currently gaining political traction. In December 2018, USDA codified three changes to the HHKFA, effective February 2019: (i) halving the whole grain mandate, (ii) halting the final sodium reduction, and (ii) allowing for flavored low-fat milk (Federal Register 2018). In March of 2019, the House committee charged with reauthorizing child nutrition programs held a public meeting discussing the need to reauthorize child nutrition programs (U.S. House Committee of Education and Labor, 2019). One month later in April, the corresponding Senate committee held their own public meeting, indicating that congresspeople are discussing reauthorization (U.S. Senate Committee on Agriculture, Nutrition, and Forestry 2019). In January of 2020, the Department of Agriculture proposed three additional changes, citing food waste and administrative burden: (i) allow for flexibilities in vegetable varieties at lunch, (ii) incorporate more a la carte options, and (iii) allow for more flexibilities in breakfast meat and grain offerings (Federal Register 2020).

While we could not investigate the leading oppositions to the HHFKA—plate waste, increased costs, and student acceptance/perceptions—a USDA-commissioned report (USDA 2019) offers some insights along these lines. First, plate waste within schools, in terms of calories, is nearly half of the national average: 21% (USDA 2019) versus nearly 40% (Hall et al. 2009). According to a systematic review (Mansfield and Savaiano 2017) there is no conclusive, nationally representative evidence as to whether plate waste increased, decreased, or remained the same as new meal standards were implemented. Second, school food authorities (the administrating entities of school-prepared food) were able to maintain their nonprofit status post-HHFKA, as they have done since the early 1990s (USDA, 2019). Third, the USDA report (USDA, 2019) also cites a high acceptance of the new school meal standards: 40% of lunch-goers stated the menu always or often includes foods they like, with another 52% stating sometimes.

A vast and robust nutritional literature on dietary habit formation has demonstrated that as few as five to ten exposures to new and novel foods can increase a child's acceptance of that food (Birch 1999; Benton 2004; Dovey et al. 2008). However, as children mature, the number of necessary exposures increases and the ability to incorporate new foods decreases. Considering that nearly half of all low-income children consume two school-prepared meals each school day, federally subsidized programs represent a substantial and repeated exposure to nutrition skill formation among the nation's most vulnerable children.

Conclusions

A primary goal of the Healthy, Hunger-Free Kids Act (HHFKA) was to improve the nutritional quality of foods consumed—not just offered or served—among students participating in the federally subsidized School Breakfast (SBP) and National School Lunch Programs (NSLP). We found the HHFKA made significant gains in the overall quality of diets: at all levels of dietary quality, not just the average, significant improvements were observed due to the consumption of a school-prepared meal in lieu of a home-prepared meal. These gains were largely due to significant improvements among middle/high school and higher-income students, although modest gains were also observed for elementary and lower-income students. In fact, among high-income, high-dietary-quality students, school meals are now of equal or better quality as compared to their home-prepared meals, rather than decreasing dietary quality as seen in the pre-HHFKA era.

What are driving the differences in the school-meal effect by grade level? Students experience increased independence in food choices as they transition into adolescence (Whitney and Rolfes 2008). Thus, older students may have more latitude in choosing foods that reflect their food preferences. Specifically, prior to the HHFKA, middle and high schools offered a broader array of foods in terms of healthfulness (Briefel et al. 2009) and older students are more likely to purchase competitive foods, which are typically less healthful (Kakarala, Keast, and Hoerr 2010). The new school meal standards changed this by requiring twice the amount of fruits and 33% more vegetables for middle and high schoolers, while further restricting the availability of competitive foods. This might partially explain why elementary school children benefited in both the pre- and post-HHFKA periods, with larger improvements concentrated among older students.

In terms of results by income level, low-income students significantly benefited from a school meal in both the pre- and post-HHKFA periods. However, the increased benefit over time (6.8% to 10.6%) was modest and statistically insignificant. Higher-income students benefited from the HHKFA to much larger degree as the school-meal effect increased from −0.8% to 4.4%. The disproportionate gain among higher-income students could be due to the following factors. First, it has been documented that lower-income students are less likely to purchase nonreimbursable competitive foods (USDA, 2019), which are typically less healthful (Kakarala, Keast, and Hoerr 2010). The HHFKA increased regulations on these foods as well, which could disproportionately affect higher-income students.¹⁰ Second, schools in low-income areas were less likely to be fully compliant with the new regulations (Au et al. 2020), again potentially attenuating the impact of the HHFKA on low-income students.

We acknowledge that we cannot fully separate the impact of the HHFKA from other structural changes that may have occurred in the postpolicy period. However, two recent studies have found nearly 75–80% of schools are fully compliant with the new standards (U.S. Department of Agriculture, Food and Nutrition Service 2019; Au et al. 2020). Thus, the main concern lies with the counterfactual diet: home-prepared food (i.e., the brown bag lunch). As we show, the quality of the at-home diet remained relatively stable over the sample period. Moreover, the quality of food children consumed at away-from-home venues (e.g., fast food and restaurants) also remained relatively stable. Thus, the results of this study provide compelling evidence that the new school meal standards provided a substantial and detectable boost in the quality of children's diets.

Finally, this study investigates the quality of foods consumed, not the quantity. The new calorie ranges could result in increased food insecurity if low-income children are not able make up any shortfalls at home. A recent study found the consumption of a school meal changes overall daily calorie intake among low- and higher-income students in the post-HHFKA period to be −54 and −124 calories, respectively (Valizadeh and Ng Forthcoming). Thus, there is some evidence that consuming a school meal does reduce calorie consumption, but how this translates into reports of food insecurity is unknown.