The spatial inequality of early care and education centers

Abstract

Given the importance of early care and education (ECE) programs for children’s development and parents’ labor force participation, it is critical to ensure communities — particularly those home to historically marginalized populations — have predictable and equitable access to programming and services. Yet, there are few useful data resources and thus relatively little research examining variation in local access to ECE programs. In this article, we examine county-level disparities in participation in and availability of ECE centers by child poverty rate, racial and ethnic composition, and urban-rural geography using a unique national dataset of county-level ECE program enrollment and expenditures from 2000 to 2019. Measures of ECE access in these data reflect a mix of publicly provided and funded programs, as well as privately-run programs that may be operated by nonprofit or for-profit entities. Findings suggest that public ECE per capita enrollment is higher in rural than urban counties, whereas private ECE program enrollment appears more highly concentrated in urban counties. Counties with higher child poverty rates also have lower enrollment rates at private ECE centers compared to counties with lower child poverty rates. We find mixed results when comparing public (e.g., public preschool, Head Start) and private ECE enrollment across counties by racial and ethnic composition. Finally, we examine year-over-year volatility in ECE enrollment and expenditures and find that public ECE programs are more stable compared to private ECE programming, which may promote equity in ECE stability.

1. Introduction

High-quality early care and education (ECE), including Head Start (HS) and Early Head Start (EHS), public or private (nonprofit or for-profit) preschool, and child care centers, has been shown to increase parental labor force participation and earnings and improve children’s economic outcomes and well-being throughout the life course (Chaudry et al., 2021; Malik et al., 2018; Morrissey, 2017). Public ECE programs such as state-sponsored preschool, HS, and child care subsidies play important roles in increasing low-income children’s access to center-based programs and, in turn, promoting health and development among low-income children (Chaudry et al., 2021; Phillips et al., 2017; Yoshikawa et al., 2013). Licensed or regulated ECE programs are often more reliable and of higher quality than unregulated options, but the high costs of private ECE lead to wide socioeconomic gaps in enrollment, which contribute to income gaps in school readiness at kindergarten (Chaudry et al., 2021; Magnuson & Waldfogel, 2016). Given the importance of ECE programs for child and parental well-being, it is critical to ensure that communities — particularly those home to historically marginalized populations — have equitable access to programming and services.

Provision of ECE, however, is an inherently local activity, reliant on the capacity of community-based, often nonprofit, organizations. Although federal funding is central to ECE in the U.S., many providers also draw upon revenues from state or local government, fee-based structures, and charitable philanthropy. Combined, these realities suggest that the presence and availability of ECE programs likely vary within and across localities. Program provision may also vary over time if public and private revenue streams fluctuate from year to year (Allard, 2009; Morrissey et al., 2022). Geographic or spatial inequality in the presence of ECE, therefore, may be an important element of how structural features of local places operate to shape child well-being and later life mobility (Morrissey et al., 2022), with important implications for equity in long- and short-term economic opportunity.

However, very little high-quality, consistently gathered, local-level ECE data exist which allow researchers to examine programs across place and time. In turn, there is relatively little research investigating spatial inequalities in ECE below the state level. To begin to fill this gap in the literature, we focus on several core research questions: how does the presence and availability of early childhood care and education centers vary within and between local places? Specifically, we examine: How does ECE provision vary by urban, suburban, or rural location, and by region of the U.S.? How does ECE provision vary by local child poverty rate? How does ECE provision vary by county racial and ethnic composition?

To answer these questions, we assemble and descriptively analyze a unique, national, county-level dataset on public and private ECE availability, expenditures, and enrollment that spans nearly two decades prior to the COVID-19 pandemic and its dramatic changes to parental labor force participation and the ECE sector (Lee & Parolin, 2021; Weiland et al., 2021). Data availability constraints mean that our study does not comprehensively capture all types of care that families use. We do not have access to administrative enrollment or expenditure data on for-profit licensed child care. Nor do we have information about regulated home-based care settings or care by relatives, which are commonly used especially for infants and toddlers. Nevertheless, >80% of children aged 3 to 5 attending weekly child care arrangements in 2019 were in center-based care, as were nearly one-third and one-half of children under 1 year and between 1 and 2 years of age, respectively (National Center for Education Statistics, 2021). Therefore, our findings are relevant to discussions about ECE policy and how resources may be better directed to increase access to care and narrow income- and race-based gaps in ECE participation.

2. Spatial variation in provision of early childhood programs

Unlike K-12 education, families bear the majority of children’s ECE costs (Chaudry et al., 2021). Over the past five decades, governments and nonprofits have expanded ECE opportunities to families across the income spectrum, often low-income households that are less likely to participate because of costs, but stand to benefit most from early education (Friedman-Krauss et al., 2023; Yoshikawa et al., 2013). First, the HS and EHS programs, created in the 1960s and 1990s, respectively, represent the federal government’s largest investment in ECE for children in poverty.¹ In addition, 44 states and the District of Columbia had some sort of public prekindergarten program in the 2021–22 academic year, which enrolled one-third of 4-year-olds and 6% of 3-year-old children with total combined funding of about $10 billion (Friedman-Krauss et al., 2023). Low-income working parents also may qualify for child care subsidies to help pay for ECE.² What results is a patchwork of ECE arrangements, with children in low-income households more often in lower-quality and informal arrangements that tend to be less reliable for working parents (Chaudry et al., 2021).

We should expect geographic variation in ECE program provision for a variety of reasons. First, the administration of ECE is highly localized. Most programs are administered by local agencies or nonprofit organizations, even if they receive federal or state dollars (Morrissey et al., 2022). For example, federal HS and EHS funding is directed to local service providers with considerable discretion in the types of programming offered.³ State- and local-sponsored public preschool programs for 3- and 4-year-olds are provided through a mixed delivery system of public schools and local nonprofits (or some for-profit entities). Local public preschool programs vary substantially in enrollment, the children they serve, and quality both within and across states, and have changed considerably over time (Friedman-Krauss et al., 2023).

Second, we expect high-poverty communities to have relatively lower ECE access. High-poverty communities may lack public resources, philanthropic resources, and nonprofit capacity necessary to provide adequate access to ECE (Allard, 2009; Morrissey et al., 2022). Moreover, many low-income families are unable to easily afford market-rate fees for private programs. Combined, these realities should lead to lower levels of ECE program supply in high-poverty places (Council of Economic Advisors, 2014; Davis et al., 2019; Malik et al., 2018). In particular, many rural communities struggle with high rates of poverty, population decline, and weakly funded public human service sectors compared to suburban or urban areas, which is associated with geographic disparities in the provision of ECE programs between rural and metropolitan areas (Allard, 2009; Davis et al., 2019; Malik et al., 2018; Morrissey et al., 2022).

Third, there is reason to expect that ECE programming will vary systematically by the racial and ethnic composition of a locality, as do many other human services. Researchers find that racial and ethnic segregation leads to access to lower quality schools, public amenities, housing, and employment opportunities (Galster & Sharkey, 2017). Prominent racial stereotypes about notions of deservingness undermine both governmental and philanthropic commitments to support low-income communities of color (Allard, 2009). The decentralized U.S. federal system allows local human service agencies to restrict access to public assistance programs in communities of color (Lieberman, 1998; Soss et al., 2011). Combined, these realities lead to human service funding that is lower, more volatile, and less responsive to need in communities of color versus predominantly White areas (Allard, 2009; Everitt & Levinson, 2016).

There is emerging evidence that delivery of ECE programs varies considerably across geography (see Morrissey et al., 2022). Research examining “child care deserts” shows stark supply differences in licensed ECE programs across community characteristics, particularly urbanicity (Malik et al., 2018; NSECE Research Team, 2016). Likewise, 6 states lacked state-funded public preschool in 2022, while 3 states had universal programs open to all children regardless of income, suggesting policy variation that may systematically affect certain types of communities (Friedman-Krauss et al., 2023). Similarly, HS and EHS availability and enrollment varies by state and community characteristics (Morrissey et al., 2022; NIEER, 2016; U.S. Government Accountability Office, 2024). There is also emerging evidence that the effects of ECE programs may vary with geographic context (Morrissey & Vinopal, 2018; U.S. Government Accountability Office, 2024).

Similarly, there is evidence of racial and ethnic disparities in ECE access. Nationally representative survey data finds that non-Hispanic White and non-Hispanic Black families are more likely to use any non-parental care when compared to Hispanic families, while non-Hispanic White children are more likely to attend center-based care than both their non-Hispanic Black and Hispanic counterparts (Cui et al., 2020). Evidence indicates that Hispanic households are more likely to live in child care deserts when compared with non-Hispanic White or non-Hispanic Black households (Malik et al., 2016). Studies have documented that both Black and Hispanic children attend lower-quality ECE programs compared to White children, disparities likely driven by variation in residential proximity to high-quality programs (McCormick et al., 2023). Notably, existing research has not studied how ECE varies across the intersection of racial and ethnic composition and geography. However, given findings on local variation in ECE access by geography and race/ethnicity, and varying histories of racial inequality across local areas, we expect that racial and ethnic inequalities in ECE access may show up differently in urban, rural, and suburban areas.

Spatial variation in ECE matters for several reasons. First, greater spatial proximity to human service programs is thought to make those programs more accessible, which should lead to higher rates of participation (Allard et al., 2003). Accessibility to critical early childhood programs may be particularly important for low-income families because of the costs imposed by complex commutes. Moreover, because ECE provides non-parental child care (Chaudry et al., 2021), geographic variability in ECE may affect parents’ economic opportunities. Such features of community context may help explain why certain areas help improve social mobility trajectories for young children more than others (Chetty et al., 2018). Indeed, research shows that the effects of HS may vary across different community contexts, with HS potentially more effective for certain child outcomes in urban vs. rural areas (McCoy et al., 2016) and in communities with few alternative ECE options (Morris et al., 2014). Thus, communities with stronger ECE infrastructure may benefit from improved outcomes that span multiple generations, including children’s health and educational attainment as well as family self-sufficiency.

Despite mounting evidence underscoring the need to advance understanding of how provision of ECE programs varies by community characteristics, most data on center-based ECE program locations, enrollment, or funding are not widely available nor standardized across local geographies. Such data limitations make it difficult to investigate local-level variation in ECE program provision. For example, while the federal Office of Head Start (OHS) provides longitudinal, annual information on HS and EHS grantees, states regulate and license child care programs and vary in whether and how those data are made public or tracked over time.⁴ These data can be interrupted by historical events, such as the pause in HS/EHS data collection in 2020 due to the COVID-19 pandemic. Moreover, individual ECE programs may receive funding from several sources (e.g., HS, state prekindergarten, and philanthropy), leading to difficulties in creating unduplicated counts of programs or enrollment (NSECE Research Team, 2014). We elaborate on these data issues and how they intersect with our analysis throughout this paper.

2.1. The current study

This study descriptively examines the geography of ECE programs by analyzing a unique, national, county-level childhood program database with multiple measures of center-based ECE participation and availability: participation in public and private preschool and nursery school programs, provision of HS/EHS program slots, and expenditures of nonprofit preschool and child care organizations. We link county-level data reflecting these center-based ECE programs with county-level Census information on urban-suburban-rural status, region, child poverty rates, and racial and ethnic demographic characteristics. In doing so, we generate descriptive insights into important structural features of local place most likely to affect variation in ECE program provision: urbanicity; poverty; and racial and ethnic composition.

3. Material and methods

Mirroring the fragmentation of ECE programs, ECE data resources are limited and inconsistently available across local or county geography. To examine local-level variation in ECE provision, this study compiles county-level data from diverse sources: American Community Survey (ACS) and decennial Census data on preschool enrollment and county demographic characteristics; administrative OHS records on HS and EHS; National Center for Education Statistics’ Common Core of Data (CCD) records on public preschool enrollment; and Internal Revenue Service (IRS) 990 form information about nonprofit “preschool” and “child day care” organization expenditures in the National Center for Charitable Statistics (NCCS). The dataset covers the years 2000 through 2019, although not all data sources are available over this whole period (see Supplementary Materials, Table A.1). Our data end just prior to the COVID-19 pandemic, which dramatically changed family life and the ECE sector starting in 2020.

Ideally, researchers would be able to merge longitudinal individual-level data to localized longitudinal ECE program data, linking individual families’ experiences and outcomes to their access to different types of ECE. In the absence of such microdata, we use counties as the unit of analysis to examine trends in ECE provision. Counties tend to be the administrative unit responsible for administering ECE programs, and in some states, child care subsidy programs. Also, institutional charitable philanthropy supporting ECE is often bounded within a given county or set of counties in a region. Prior research finds that relatively large geographic areas such as counties can accurately capture the markets for child care centers (Gordon & Chase-Lansdale, 2001). More practically, it is difficult to obtain reliable and comparable data on ECE and community characteristics at smaller units of geography.⁵

We distinguish between public and private ECE programming to the extent possible given available information. ECE program data provides limited insight into public versus private provision, despite the importance of these distinctions. We consider public programs as those which are predominantly funded and regulated or administered by public entities, such as state-sponsored preschool, preschool within public school settings, and HS/EHS. We consider private provision to be programs administered by non-governmental, private entities (e.g., nonprofit organizations) or programs that ACS respondents self-report as “private,” which may occur through nonprofit or for-profit entities. We acknowledge that distinctions between public and private in ECE may be blurred in our data (as they are in practice).⁶ Despite these complexities, we believe there is utility in differentiating between public and private ECE programs and are cautious in our interpretations given the realities of ECE data and practice.

3.1. County-level ECE data

This study combines several county-level sources of ECE participation and availability.⁷ First, ACS data aggregates parental reports of preschool and nursery school enrollment of 3- and 4-year-olds, including whether a child attended a public or private option (the latter including both nonprofit and for-profit status).

Second, annual data from OHS Program Information Reports (PIRs) between 2000 and 2019 report funded enrollment for all HS and EHS programs, linked to the county location of the program’s administrative headquarters. PIR data contain rich annual information on HS and EHS programs, but connect to the location of the administrative headquarters and do not reflect instances where programs may be operating HS or EHS centers in multiple counties. As a result, we also present data from OHS center-level records that include more granular program location information, but are only available in 2013, 2014, 2015, and 2019. Discussion of findings below makes explicit reference to “HS/EHS program enrollment,” when program-level data are being analyzed, or to “HS/EHS center slots,” when examining center-level data. To create county-level measures of programming relative to potential demand, we divide EHS program slots by the number of children under three with household income at or below 150% of the federal poverty level (FPL), and HS program slots by the number of 3- and 4-year-olds with household income at or below 150% FPL.⁸

Third, we aggregate local public school pre-K enrollment data reported in the Common Core of Data (CCD) to the county level for years 2000 to 2019. The CCD contains detailed information on all schools operated by public school districts across the United States. We report the number of pre-K students enrolled per 100 children under 5 in each county.⁹ We refer to these pre-K enrollment data from the CCD as “public pre-K” throughout this paper.

Finally, we use IRS form 990 filings from 2000 to 2019 to examine inflation-adjusted nonprofit expenditures per child under five years of age by organizations self-identifying as preschool or day care service nonprofits.

In addition to reporting these measures of ECE provision per capita in a given year, we also examine how ECE enrollment and provision changes within a given county over time with two measures of year-over-year volatility. First, we use the arc percent change (APC) in each per capita metric to capture year-over-year variability. We report the standard deviation (SD) across all counties’ year-over-year APC in ECE provision, which summarizes year-over-year volatility experienced across all counties in a given year.¹⁰ Second, we report year-to-year changes in ECE availability by magnitude and direction. We report what share of changes were “large negative” changes (negative APC of >10%), “small” changes (APC between −10% and 10%) or “large positive” changes (positive APC >10%).

3.2. County-level geographic identifiers

Counties are sorted into urban, suburban, and rural categories based the Office of Management and Budget (OMB) definitions of metropolitan and non-metropolitan area boundaries and urban-rural continuum codes (USDA, 2019). “Urban” counties contain the primary urban center of a metropolitan area. “Suburban” counties are part of the same metropolitan area as urban counties but do not contain the metro’s primary city. “Rural” counties are non-metropolitan regions as defined by OMB.¹¹ We also disaggregate by U.S. census region: Northeast, South, Midwest, and West.

3.3. County-level demographics

Child poverty rates are drawn from the 2017–2021 ACS five-year estimates, which reflect estimates of the percent of children under 18 living in households below the federal poverty threshold. We group counties by whether they have a child poverty rate <10% (low poverty), between 10% and 20% (moderate poverty), and 20% or greater (high poverty).¹² We also examine how ECE program provision varies across counties where children of color compose a larger versus a smaller share of the population. Because racial and ethnic diversity varies widely across urban, suburban, and rural counties, however, it is difficult to create a single threshold reflecting racial or ethnic population densities that could be applied to all types of geography. Within urban, suburban, and rural county categories, therefore, we use the ACS to identify counties that had a high (low) share Black or Hispanic under-5 population as counties in which the percent of children under age 5 identifying as Black or Hispanic was more than half a standard deviation above (below) the mean within that county geography.¹³

3.4. Limitations of available ECE data

While these data provide important insights into the geography of ECE access, results should be interpreted with several caveats. First, self-reported survey data may provide under- or over-estimates of enrollment in public or private ECE programs due to respondent errors or lack of knowledge. Administrative data could provide more accurate snapshots of caseloads and program output, but such data can be limited in scope and subject to changing data collection and reporting requirements.

Not all our ECE provision and participation measures are mutually exclusive; data limitations result in some duplication across measures. For example, there is modest overlap between enrollment slots reported in the CCD and the number of 3- and 4-year-olds in public preschool reported by ACS and in HS/EHS data.¹⁴ This overlap complicates our efforts to assess public preschool provision over time. Furthermore, HS services may be delivered by nonprofit (or for profit) entities, thus appearing in counts of HS services as well as tallies of nonprofit service expenditures. We are cautious in our interpretations accordingly.

Further, our data lack information on regulated home-based ECE settings, which inhibits our ability to capture the full ECE landscape. High-quality, longitudinal information on home-based programs is extremely sparse. Unfortunately, child care licensure data are maintained by states in different ways and for various years, and thus our dataset includes center-based programs only. Given the importance of home-based care for children from racial and ethnic minority backgrounds and immigrants (Meek et al., 2020) and infants and toddlers (Cui et al., 2020), and differences in quality between center- and home-based settings (Chaudry et al., 2021), exploring access to home-based ECE remains an important area for future research. Relatedly, this study lacks administrative data on private for-profit programs, which include most family child care programs and an estimated 30% of centers in 2019 (Datta et al., 2021).¹⁵ However, we capture a subset of private for-profit center-based programs in ACS survey data on private “preschool or nursery school” and in EHS/HS data (as some for-profits deliver EHS/HS).

Readers should keep in mind that ECE programs in our dataset are targeted at children of different ages. Some of our data sources (e.g., the nonprofit data) describe programs that may be attended by children under 5 of any age, while others (such as Head Start) target preschoolers specifically. We adjust the denominator in each of our ECE metrics to match the children targeted by the program as closely as possible, but results must be interpreted acknowledging that the age groups served vary across programs. ACS survey data only capture preschool attendance among 3- and 4-year-olds, which is a significant limitation especially given that roughly half of children under 3 participate in a non-parental care arrangement (Cui et al., 2020).

Furthermore, county-level data smooth over community and neighborhood variation in ECE program provision that likely affects ECE participation. For example, research using high-quality microdata from a single locality has documented racial and ethnic disparities in ECE attendance within local areas (McCormick et al., 2023). While we compare ECE program attendance across counties by child poverty rates and racial composition, we do not capture ECE participation or access among individual children in poor households or children of color.

Finally, data from the NCCS have several limitations. These data reflect only nonprofit organizations that submit 990 forms to the IRS. Nonprofits self-report one core service code, even though many organizations could fit into multiple service categories. IRS 990 forms also do not break out expenditures or revenues by program or client population. Nonprofit data from the IRS, like EHS and HS PIR data, only contain location information about an organization’s administrative headquarters and not separate offices where services may be delivered. IRS data, therefore, may misrepresent large social service nonprofits that operate programs in rural regions or suburban communities, but maintain headquarters in a central city or rural population center. Finally, findings on expenditures do not necessarily reflect differences in quality or quantity of ECE services provided given differences in local costs of providing ECE.

Given the limitations of available ECE program data, we believe our measures reflect rigorous measurement of ECE access as discussed in the ECE and human services research literatures. First, when possible, we are careful to account for service supply and potential demand in creating measures of service provision and access (see Allard, 2009). Second, we include measures of access that reflect the number of slots available and children enrolled, which are consistent with definitions of access that reflect the “reasonable effort” families might have to make to participate in ECE (Friese et al., 2017). Distinctions between privately and publicly funded or provided ECE programming also touch upon affordability dimensions of access, given that costs of publicly provided ECE are often partially or fully subsidized (Friese et al., 2017). We examine both the stock and flow of ECE programming, which is critical to understanding levels and volatilities in access (Allard & Pelletier, 2023). Finally, we analyze ECE access with attention to the experiences of historically marginalized communities and at-risk children.

4. Results

Below, we summarize our findings regarding variation in ECE access by key county-level characteristics: geography, poverty, and race/ethnicity.

4.1. National trends in ECE provision

First, we situate our findings within the nationwide landscape of public and private ECE provision. Fig. 1 shows trends in each of our main ECE metrics over time in all U.S. counties. Panel A indicates that despite increased public discussion about the importance of high-quality early childhood education, the number of 3- and 4-year-olds enrolled in preschool and nursery school remained relatively constant between 2007 and 2019. The number of 3- and 4-year-olds enrolled in school fell 5.2% (from 5.0 million to 4.7 million), while the enrollment rate fell 2 percentage points (from 47% to 45% – not shown in Fig. 1).¹⁶ The share of enrolled children who attended public preschool or nursery school, however, rose by 1.4 percentage points (from 25.2% to 26.6%), suggesting that public programs comprised a slightly larger share of ECE enrollment as the overall number of children enrolled fell slightly.

Fig. 1. Nationwide trends in ECE provision.

Notes: Head Start and Early Head Start enrollment figures in this graph refer to funded enrollment at the program level. Nonprofit expenditures are adjusted for inflation and reported in 2019 dollars. Includes NTEE codes B21 (Education: Elementary and Secondary Schools: Preschools) and P33 (Children & Youth Services: Child Day Care). Panels have different time windows reported due to differences in data availability.

Sources: ACS 5-year estimates 2017–21; OHS; CCD; NCCS

There also are interesting dynamics in enrollment across the major public ECE programs over time. Panel B traces enrollment across three major public ECE programs today: HS, EHS, and state-funded public pre-K slots delivered through traditional public schools.¹⁷ Overall enrollment in these core types of ECE programming rose 38% between 2002 and 2019 (from 1.6 million to 2.2 million slots). EHS, though a small share of overall ECE slots, expanded rapidly during this period, providing 261% more slots in 2019 compared to 2002 (from 46,325 slots to 167,261). Given the small relative size of EHS, it appears that enrollment in public pre-K programs through traditional public schools fueled much of the overall growth in ECE participation since 2002. Between 2002 and 2019, public school districts expanded ECE by roughly 600,000 slots – an increase of 29% (from 756,350 to 1,356,591). Finally, Panel C shows that nationwide spending by nonprofit preschool and day care organizations rose between 2000 and 2004, from just over $6 billion to over $9 billion, then levelled off through 2019; in 2019, these organizations spent a total of roughly $8 billion.

4.2. Examining variation in county-level ECE provision

Aggregated national trends obscure disparities in ECE participation and access across local contexts. Tables 1 and 2 report ECE program participation and provision metrics in 2019 – the most recent year where all data are reliably available – disaggregated by county geography, child poverty rate, and region.¹⁸

Table 1.

Early care and education enrollment and nonprofit expenditures by county geography, region, and child poverty rate, 2019.

	County Geography			Region				Child Poverty Rate
	Urban	Suburban	Rural	Northeast	South	Midwest	West	0–10%	10–20%	20%+
A. % of 3–4 year olds enrolled in preschool or nursery school
Mean	44.2a	43.2b	41.2ab	47.6abc	41.4a	42.1b	41.2c	45.2ab	41.5a	41.3b
Median	43.7	43.1	40.3	47.9	40.7	42.2	40.5	45.6	41.8	40.1
SD	8.9	13.3	17.1	11.7	16.2	14.3	17.2	17.6	13.6	16.5
N	383	716	2037	217	1421	1054	439	511	1331	1294
B. Of enrolled 3–4 year olds, % enrolled in private school
Mean	37.8a	36.0b	21.6ab	37.5a	26.3a	23.7a	31.1a	32.8a	30.2a	21.1a
Median	38.8	36.5	18.8	38.4	23.7	22.3	30.4	33.9	29.2	18.9
SD	12.8	18.8	18.7	15.3	20.1	16.5	22.8	22.5	18.7	17.3
N	383	716	2037	217	1421	1054	439	511	1331	1294
C. Expenditures by nonprofit preschool and daycare organizations per child under 5
Mean	430.4ab	195.1a	238.6b	653.2ab	166.7ac	202.9bd	447.2cd	398.6ab	251.1a	195.2b
Median	189.3	0.0	0.0	221.5	0.0	0.0	0.0	0.0	0.0	0.0
% $0	13.0%	60.1%	72.4%	22.1%	72.2%	60.9%	53.8%	55.4%	55.2%	72.4%
SD	749.6	837.5	1116.8	1739.3	851.9	650.9	1566.6	1199.0	631.7	1241.1
N	383	716	2037	217	1421	1054	439	511	1331	1294

Notes: Letter annotations (e.g., a, b) indicate that the difference of means for within row pairs are statistically significant from zero at 0.10 level; letters are bolded if p<0.05 and unbolded if 0.05 < p < 0.10. Nonprofit expenditures are adjusted for inflation and reported in 2019 dollars.

Sources: ACS 5-year estimates 2017–21; OHS; USDA ERS; OMB; CCD; NCCS.

Table 2.

Public early care and education enrollment by county geography, region, and child poverty rate, 2019.

	County Geography			Region				Child Poverty Rate
	Urban	Suburban	Rural	Northeast	South	Midwest	West	0–10%	10–20%	20%+
A. Public pre-K enrollment per 100 children under 5 years old
Mean	8.3a	10.3a	14.6a	8.7ab	13.4ac	13.7bd	10.5cd	13.0	11.9a	13.7a
Median	7.5	9.1	13.1	7.5	12.1	12.5	7.3	9.8	10.3	12.5
SD	5.2	6.6	10.4	7.4	8.2	9.6	13.2	11.5	8.6	9.2
N	343	678	1861	199	1388	958	332	475	1202	1205
B. Head Start program enrollment per 100 poor children aged 3–4
Mean	38.4	25.1a	42.2a	39.5	31.9	40.2	50.9	39.7	37.6	37.4
Median	32.1	0.0	0.0	31.4	0.0	0.0	0.0	0.0	0.0	0.0
SD	36.9	63.9	170.5	50.0	97.1	158.2	225.8	194.8	131.2	126.3
N	383	715	2021	217	1412	1051	434	499	1329	1291
C. Head Start center slots per 100 poor children aged 3–4
Mean	34.1a	40.0b	61.0ab	45.1	53.3	51.5	59.2	53.0	48.7a	57.1a
Median	32.2	29.9	45.2	40.7	36.5	39.8	40.1	34.9	37.9	41.0
SD	18.7	57.2	109.9	25.2	119.6	49.2	100.4	78.5	58.0	123.6
N	383	715	2021	217	1412	1051	434	499	1329	1291
D. Early Head Start program enrollment per 100 poor children under 3
Mean	6.0	4.3a	8.3a	7.5	4.5a	9.9a	8.7	11.4a	7.3	5.3a
Median	4.1	0.0	0.0	2.8	0.0	0.0	0.0	0.0	0.0	0.0
SD	7.7	12.9	51.3	11.5	21.7	62.0	40.3	82.4	32.3	22.9
N	383	715	2021	217	1412	1051	434	499	1329	1291
E. Early Head Start center slots per 100 poor children under 3
Mean	3.4a	3.7b	6.3ab	6.4	4.0a	6.6a	6.5	6.3	5.4	5.0
Median	2.7	0.0	0.0	3.4	0.0	0.0	0.0	0.0	0.0	0.0
SD	3.5	6.7	23.0	8.9	23.6	13.3	16.4	16.6	12.1	24.6
N	383	715	2021	217	1412	1051	434	499	1329	1291

Notes: Letter annotations (e.g., a, b) indicate that the difference of means for within row pairs are statistically significant from zero at 0.10 level; letters are bolded if p<0.05 and unbolded if 0.05 < p < 0.10. Poor children are children living in a household with income < 150% FPL.

Sources: ACS 5-year estimates 2017–21; OHS; USDA ERS; OMB; CCD.

4.3. Urban, suburban, and rural counties

Rural counties averaged lower overall preschool enrollment rates than urban and suburban localities, although patterns differed by ECE type. For example, whereas nearly 40% of preschoolers in urban and suburban counties attended a private preschool, on average, less than one-quarter of preschoolers in rural counties attended a private school (see Row B in Table 1). Average spending by nonprofit preschool and day care organizations somewhat mirrors this finding; ECE nonprofit organizations located in urban counties spent 2.2 and 1.8 times more per child under 5 years of age on average compared to those located in suburban and rural counties ($430 versus $195 and $239, respectively – see Row C of Table 1).

While private ECE programming was more present in urban counties, public ECE programs played a larger role in rural areas. For example, enrollment in pre-K through the public school system was higher in rural counties relative to suburban and urban counties. Nearly 15 of every 100 children under age 5 in the mean rural county attended a public pre-K program compared to 10.5 per 100 children under 5 in suburban counties and 8.5 per 100 children under 5 in urban counties (see Row A in Table 2). HS and EHS play a similarly larger role in the rural ECE landscape. While urban counties had an average of 38 slots in HS centers per 100 children with income <150% FPL, rural counties had nearly double that, with an average of 61 slots per potential attendee (see Row B in Table 2).

4.4. U.S. regions

The middle columns of Tables 1 and 2 also demonstrate variation in both public and private provision across regions of the country. Table 1 shows that counties in the Northeast had slightly higher overall preschool enrollment rates on average (47.6%) compared to counties in the South (41.4%), Midwest (42.1%), and West (41.2%). Private preschool enrollment was more prevalent, and nonprofit ECE spending higher, in counties across the Northeast and West regions of the country than in the South or Midwest. The average county in the Northeast was home to nonprofit preschool and day care organizations that spent $653 per child under five years old – spending that was roughly three to four times higher than observed in the average county in the Midwest or South (see Table 1, Row C). Furthermore, a large percentage of counties outside the Northeast have no registered nonprofit preschool and day care organizations, as evidenced by the fact that the median county in the South, Midwest, and West recorded $0 in expenditures by such nonprofit ECE organizations.

More modest differences in public ECE provision are apparent across regions. For example, Table 2 shows that average public pre-K enrollment was slightly higher in the South (13.4 slots/100 children under 5) and Midwest (14.1 slots/100 children under 5) compared to the West (11.6 slots/100 children under 5) and Northeast (8.9 slots/100 children under 5) (see Table 2, Row A). Surprisingly, counties in the South consistently had fewer EHS slots per low-income children than other regions. Counties in the southern U.S. averaged 4 allocated EHS slots per 100 children with income <150% FPL, while counties across all other regions had between 6.4 and 6.6 slots per capita (see Table 2, Row D).

4.5. Child poverty rates

Consistent with prior research looking at tract-level ECE attendance (Morrissey & Vinopal, 2018), county-level analyses show that ECE participation was lower in counties with higher child poverty rates. In counties with high child poverty rates (over 20%), 41.3% of 3- and 4-year-old children were enrolled in preschool in 2019 on average, compared to 45.2% in counties with low child poverty rates (<10%) (see Table 1, row A). Counties with lower child poverty rates have a relatively larger share of children enrolled in private preschools and nursery schools relative to public ones. An average of nearly one-third of children enrolled in preschool attended a private preschool in low-poverty counties, compared to 21.1% of children in high-poverty counties (see Table 1, Row B). Average per capita expenditures by nonprofit preschool and day care organizations was also lower in higher-poverty counties. Counties with low child poverty rates were home to nonprofit preschool and day care programs that spent an average of $400 per child under five annually, compared to programs that spent $195 per child under five in counties with high child poverty rates (see Table 1, Row C).

We find mixed evidence as to whether there are disparities in public ECE provision across low-, moderate-, and high-poverty counties. Enrollment in public pre-K was relatively similar across counties with low and high child poverty rates, although counties with moderate child poverty rates (between 10% and 20%) had slightly less public pre-K enrollment (see Table 2, Row A). HS enrollment was relatively consistent across counties with lower versus higher child poverty rates when we examine program-level data (Row B). Center-level data, however, suggest that high-poverty counties had significantly more slots in HS centers on average (57.1 per 100 poor children) compared to moderate-poverty counties (with 48.7 slots per 100 poor children) and slightly more slots per child than in low-poverty counties (with 53.0 slots per 100 poor children), even after adjusting for the number of poor children in a county (Row C).

Trends in ECE provision and county child poverty rates are reversed, however, when we examine EHS slots. Program-level data on EHS indicates that programs headquartered in high-poverty counties operated only 5.2 EHS program slots per 100 children with income <150% FPL compared to 11.6 in the lowest-poverty counties (see Table 2, Row D). When we examine EHS center-level data, however, these differences are smaller and no longer statistically significant, suggesting that the location of ECE centers is more equitable than program location data would suggest (see Table 2, Row E). Findings reported in Tables 1 and 2 are somewhat consistent with a U-shape participation curve commonly observed, where middle-income children cannot afford private ECE programs, but are not eligible for public ECE options.¹⁹

4.6. County racial and ethnic composition

Tables 3–5 analyze how ECE program metrics vary across county racial and ethnic composition. Among urban, suburban, and rural counties, we examine how ECE attendance and provision varies between counties with a relatively low (more than half a standard deviation below the mean) or high (more than half a standard deviation above the mean) share of the population under 5 identifying as Black or Hispanic. Counties where Black children composed a larger share of the population averaged higher percentages of 3- and 4-year-olds enrolled in preschool (46.0% to 47.9%, see Row A of Table 3), compared to counties where Black children composed a smaller share of the population (40.1% to 41.5%). Large differences in rates of private preschool attendance exist between urban and suburban counties where Black children composed a large share of the population (36.2% and 37.9%, respectively; see Table 3, Row B) and rural counties with larger shares of Black children (20.9%). Private preschool attendance was higher within suburban counties where Black children composed a larger share of the population than suburban counties where Black children were a smaller share of the population.

Table 3.

Early care and education enrollment and nonprofit expenditures by county geography and racial/ethnic composition, 2019.

	County Geography
	Urban				Suburban				Rural
	Mean	Med.	SD	N	Mean	Med.	SD	N	Mean	Med.	SD	N
A. % of 3–4 year olds enrolled in preschool or nursery school
Share Black population under 5
Low	41.5a	40.5	8.4	153	40.1ab	39.4	12.9	309	-	-	-	-
Moderate	44.8a	44.0	9.0	138	44.6a	44.9	12.1	275	40.3a	39.7	16.9	1722
High	47.9a	46.7	8.1	92	47.4b	45.7	14.9	132	46.0a	43.9	17.9	315
Share Hispanic population under 5
Low	45.6a	45.6	8.4	157	41.9a	41.3	15.1	264	40.9	39.6	18.0	767
Moderate	43.8	42.6	9.4	142	43.5	43.0	12.1	313	41.7	41.4	15.2	902
High	42.5a	40.4	8.8	84	44.9a	45.7	11.9	139	40.8	39.1	19.6	386
B. Of enrolled 3–4 year olds, % enrolled in private school
Share Black population under 5
Low	37.9	38.7	13.7	153	30.6ab	28.4	18.7	309	-	-	-	-
Moderate	39.0	39.2	11.9	138	41.3a	42.5	16.5	275	21.7	19.2	18.8	1722
High	36.2	38.2	12.5	92	37.9b	35.4	19.9	132	20.9	16.8	18.0	315
Share Hispanic population under 5
Low	36.4a	36.9	11.9	157	30.6ab	29.1	18.8	264	20.0a	16.2	18.9	767
Moderate	42.4ab	42.7	12.2	142	38.1a	39.0	18.0	313	23.6ab	21.1	17.9	902
High	32.9b	33.4	13.1	84	41.7b	43.9	18.0	139	19.9b	16.7	19.5	368
C. Expenditures by nonprofit preschool and day care organizations per child under 5
Share Black population under 5
Low	432.8	187.4	783.1	153	131.7	0.0	389.8	309	-	-	-	-
Moderate	418.8	210.7	707.3	138	235.5	11.9	629.9	275	258.9a	0.0	1132.5	1722
High	443.9	156.5	762.4	92	259.5	0.0	1620.0	132	127.2a	0.0	1021.2	315
Share Hispanic population under 5
Low	374.7	169.1	700.2	157	110.0	0.0	392.7	264	254.5	0.0	1348.7	767
Mod.	425.3	212.8	540.9	142	254.4	0.0	1173.2	313	241.4	0.0	1068.2	902
High	543.2	178.9	1072.0	84	223.2	18.2	453.9	139	198.5	0.0	565.3	368

Notes: A county was considered to have a low (high) percentage Hispanic or Black population under 5 if the share of the county’s population under 5 identifying as Hispanic or Black was more than half a standard deviation below (above) the mean across all suburban or urban counties in a given year. Letter annotations (e.g., a, b) indicate that the difference of means for within column pairs are statistically significant from zero at 0.10 level; letters are bolded if p<0.05 and unbolded if 0.05 < p < 0.10.

Sources: ACS 5-year estimates 2017–21; USDA ERS; OMB; CCD; NCCS

Table 5.

Early Head Start enrollment by county geography and racial/ethnic composition, 2019.

	County Geography
	Urban				Suburban				Rural
	Mean	Med.	SD	N	Mean	Med.	SD	N	Mean	Med.	SD	N
A. Early Head Start program enrollment per 100 poor children under 3
Share Black population under 5
Low	6.9a	4.5	10.0	153	4.7	0.0	15.7	309	-	-	-	-
Mod.	4.9a	3.5	5.3	138	4.5	0.0	10.9	274	9.0	0.0	55.2	1706
High	6.0	4.5	6.0	92	2.7	0.0	8.7	132	4.5	0.0	18.1	315
Share Hispanic population under 5
Low	6.1	4.5	8.1	157	4.4	0.0	16.0	263	9.5	0.0	67.5	758
Mod.	5.6	4.0	5.8	142	4.3	0.0	12.0	313	8.4	0.0	43.1	901
High	6.5	3.3	9.6	84	4.0	0.0	7.1	139	5.7	0.0	22.6	362
B. Early Head Start center slots per 100 poor children under 3
Share Black population under 5
Low	4.2ab	2.9	4.4	153	4.8ab	0.0	8.2	309	-	-	-	-
Mod.	3.2a	2.6	2.9	138	2.9a	0.2	4.9	274	6.8a	0.0	24.7	1706
High	2.6b	2.1	2.4	92	2.9b	0.0	5.8	132	3.9a	0.0	9.6	315
Share Hispanic population under 5
Low	3.5	2.9	3.7	157	4.7ab	0.0	8.5	263	6.3	0.0	15.7	758
Mod.	3.5	2.8	3.4	142	3.3a	0.0	5.8	313	5.8	0.0	11.9	901
High	3.3	2.2	3.5	84	3.0b	1.3	4.5	139	7.8	0.0	45.7	362

Notes: A county was considered to have a low (high) percentage Hispanic or Black population under 5 if the share of the county’s population under 5 identifying as Hispanic or Black was more than half a standard deviation below (above) the mean across all suburban or urban counties in a given year. Letter annotations (e.g., a, b) indicate that the difference of means for within column pairs are statistically significant from zero at 0.10 level; letters are bolded if p<0.05 and unbolded if 0.05 < p < 0.10. Nonprofit expenditures are adjusted for inflation and reported in 2019 dollars. Poor children are children living in a household with income <150% FPL.

Sources: Office of Head Start; ACS 5-year estimates 2017–21; USDA ERS; OMB.

Table 3 provides evidence of similar, but smaller, gaps in preschool attendance between suburban counties where Hispanic children composed a larger or smaller share of the population under 5. Urban counties with larger Hispanic populations had slightly lower preschool enrollment rates, on average, than urban counties with smaller Hispanic populations. These findings mirror prior research demonstrating that Hispanic families are more likely to live in child care deserts than their non-Hispanic counterparts (Malik et al., 2016). Urban and rural counties with the highest share of Hispanic children had slightly less participation in private preschool, on average. In contrast, suburban counties with a higher share of Hispanic children had a higher share of preschoolers enrolled in private preschool.

Consistent with findings in Table 1, Table 3 indicates that per capita nonprofit preschool and day care expenditures were much higher in urban counties than suburban or rural counties, regardless of racial and ethnic composition. Rural counties with a relatively higher share of Black or Hispanic children, however, averaged lower nonprofit ECE expenditures than rural counties where a smaller share of the population was Black or Hispanic. Although the differences are modest, suburban counties where Black or Hispanic children composed a higher share of the population had slightly higher nonprofit ECE expenditures per capita on average than suburban counties where a lower percentage of the population was Black or Hispanic.

Table 4 suggests there is little variation in public pre-K provision across urban counties where Black children composed a smaller or larger share of the population (see Table 4, Row A). Suburban and rural counties where Black children composed a higher share of the population, however, had significantly fewer public pre-K slots than suburban and rural counties where Black children composed a smaller share of the population. These differences are more pronounced in rural areas; counties where Black children composed a higher share of the population have 5 fewer public pre-K slots per 100 children compared to counties where Black children comprised a lower share of the population. Table 4 provides mixed results regarding the share of the population comprised of Hispanic children (see Row A). Urban and rural places with a relatively large share of Hispanic children had slightly more public pre-K slots per capita compared to comparable areas with a smaller share of Hispanic children. Suburban places with a relatively larger share of Hispanic children had slightly fewer public pre-K slots.

Table 4.

Public pre-K and Head Start enrollment by county geography and racial/ethnic composition, 2019.

	County Geography
	Urban				Suburban				Rural
	Mean	Med.	SD	N	Mean	Med.	SD	N	Mean	Med.	SD	N
A. Public pre-K enrollment per 100 children under 5 years old
Share Black population under 5
Low	8.8	7.7	5.9	122	11.6ab	10.2	7.0	285	-	-	-	-
Moderate	8.1	7.5	4.6	129	9.0a	8.1	6.1	265	15.0a	13.8	10.9	1562
High	7.9	7.0	5.0	92	10.0b	9.4	6.0	128	12.3a	11.4	6.7	299
Share Hispanic population under 5
Low	8.5	8.0	5.0	151	11.7ab	10.2	7.6	257	14.8ab	12.8	12.2	694
Moderate	7.4a	6.4	5.3	132	9.3a	8.2	5.8	298	13.5a	12.7	8.4	825
High	10.1a	10.2	5.0	60	9.7b	9.2	5.3	123	16.8b	16.1	10.5	342
B. Head Start program enrollment per 100 poor children aged 3–4
Share Black population under 5
Low	43.9a	36.1	42.3	153	24.5	0.0	68.1	309	-	-	-	-
Moderate	31.9a	29.0	29.1	138	25.4	0.0	51.1	274	43.0	0.0	180.8	1706
High	39.0	34.9	36.6	92	26.1	0.0	76.7	132	38.0	0.0	98.1	315
Share Hispanic population under 5
Low	40.8	37.0	36.3	157	24.9	0.0	71.1	263	44.2	0.0	190.1	758
Moderate	33.6	28.5	32.4	142	25.3	0.0	66.4	313	41.3	0.0	133.6	901
High	42.1	29.2	44.0	84	25.1	11.3	39.4	139	40.5	0.0	206.1	362
C. Head Start center slots per 100 poor children aged 3–4
Share Black population under 5
Low	38.0a	36.0	19.3	153	48.2a	36.6	74.5	309	-	-	-	-
Moderate	30.4a	29.2	14.5	138	29.9a	25.1	25.0	274	59.5	43.6	109.8	1706
High	33.1	30.0	21.9	92	41.6	27.1	56.1	132	69.3	51.4	110.7	315
Share Hispanic population under 5
Low	36.3	34.1	18.6	157	50.1ab	38.3	66.3	263	64.0	49.6	87.2	758
Moderate	31.7	29.3	18.7	142	35.3a	26.6	58.2	313	53.6a	42.8	50.1	901
High	34.1	31.3	18.6	84	31.2b	24.5	24.9	139	73.1a	40.9	212.4	362

Notes: A county was considered to have a low (high) percentage Hispanic or Black population under 5 if the share of the county’s population under 5 identifying as Hispanic or Black was more than half a standard deviation below (above) the mean across all suburban or urban counties in a given year. Letter annotations (e.g., a, b) indicate that the difference of means for within column pairs are statistically significant from zero at 0.10 level; letters are bolded if p<0.05 and unbolded if 0.05 < p < 0.10. Nonprofit expenditures are adjusted for inflation and reported in 2019 dollars. Poor children are children living in a household with income <150% FPL.

Sources: Office of Head Start; ACS 5-year estimates 2017–21; USDA ERS; OMB.

Urban and suburban counties where Black and Hispanic children composed a larger share of the population had fewer slots in HS centers per capita (see Table 4, Row C). For example, there were 38.0 slots in HS centers per 100 poor children three to four years old on average in urban counties where Black children composed a relatively small share of the population, compared to 33.1 HS center slots per 100 poor children 3 to 4 years old where Black children composed a relatively larger share of the population. Suburban counties where Hispanic children composed a relatively small share of the population were home to 50.1 HS center slots per 100 poor children 3 to 4 years old on average, compared to 31.2 HS center slots per 100 poor children 3 to 4 years old on average in suburban counties where Hispanic children composed a larger share of the population. In rural counties, we find evidence that those with relatively larger Black and Hispanic populations had more HS center slots, on average, than counties with relatively smaller Black and Hispanic populations. Similar trends are present for EHS enrollment with one exception: rural counties with a larger share of Black children have fewer EHS center slots, on average, than those with a smaller share of Black children (see Table 5).²⁰

4.7. Volatility over time

Fig. 2 reports year-over-year volatility in key county-level ECE measures, reflecting the extent to which ECE provision fluctuated within counties over time.²¹ Panels A and B display year-over-year volatility in ECE provision as measured by the standard deviation of the year-to-year APC in each ECE metric. This measure reflects the extent to which ECE enrollment or access varied across all counties. Panel A displays year-over-year variability in public ECE programs, as measured by the enrollment rate of 3- and 4-year-olds in public preschool and per capita funded enrollment for public pre-K, HS, and EHS programs. In Panel B, volatility in private ECE is represented by the enrollment rate of 3- and 4-year-olds in private preschool and per capita nonprofit preschool and day care expenditures. Fig. 2 provides evidence that public ECE provision was typically less variable within counties when compared to private ECE provision. One exception is spikes in volatility of EHS provision in 2010 and 2017, when the program was expanded substantially. Such increases likely reflect the severe economic hardship and the subsequent injection of government stimulus supporting ECE following the Great Recession (e.g., the American Recovery and Reinvestment Act of 2009), which increased both family eligibility and resources available for EHS.

Fig. 2. Volatility in year-over-year early care and education enrollment and nonprofit expenditures.

Notes. A large negative (positive) change in an ECE provision metric was defined as a year-over-year arc percent change <−10% (>10%). A small change was a year-over-year arc percent change between −10% and 10%. Nonprofit expenditures are adjusted for inflation and reported in 2019 dollars. Poor children are children living in a household with income <150% FPL.

Sources: ACS 5-year estimates 2017–21; OHS; USDA ERS; OMB; CCD; NCCS.

Panel B breaks down year-over-year changes in ECE measures by change size and directionality. Results highlight significant variability in ECE participation measures within counties over time. More than half of counties experienced large positive or negative changes in enrollment in public and private ECE programs, but differences between public and private programs emerge. Not only was enrollment in private preschool programs more volatile than enrollment in public programs, but a larger share of counties experienced sizeable reductions in private enrollment from one year to the next compared to public enrollment. From any one year to the next, 34% of counties experienced a large reduction in enrollment in private preschool, while 27% of counties experienced a similarly large reduction in public enrollment.

Consistent with a policy context supportive of public ECE program expansion in recent years, counties experienced large increases in public pre-K and EHS. This was also true for HS enrollment, although HS was more stable than either public pre-K or EHS. Roughly half of counties experienced large changes in nonprofit ECE spending in any given year, with about 24% experiencing large negative changes and 24% experiencing large positive changes. Together, these findings suggest that public ECE programs, when compared to private ECE programs, were more stable and relatively more likely to experience expansion.

5. Discussion

The accessibility of affordable, high-quality ECE programming is critical to parents’ labor force participation and promoting children’s educational and other outcomes (e.g., Chaudry et al., 2021). This study finds evidence of geographic inequities in ECE participation and program investments. While we believe these findings are relevant to policy research discussions around ECE programming and the provision of center-based care, it is important to bear in mind that a majority of families in the U.S. rely on in-home care for infants and toddlers (National Center for Education Statistics, 2021).

In general, results indicate that private ECE programs have lower average enrollment, less availability, and more volatility over time, particularly in rural communities. Nonprofit ECE expenditures also are much lower in rural communities compared to urban and suburban areas. While expenditure data provide blunt indicators of provision, particularly given geographic variation in the cost of living, expenditure levels do present insight into resources available that may shape program availability or quality. Such patterns are consistent with research finding that rural communities are more likely to lack child care services (Malik et al., 2018), likely because lower levels of income, lower maternal employment rates, and lower economies of scale may weaken supply and demand in rural areas (Morrissey et al., 2022; Ziliak, 2019).

We also find that public ECE programs, particularly HS, play a particularly important role in providing early learning opportunities to young children living in rural areas relative to those in metropolitan areas. Such detailed descriptive results echo findings of geographic disparities in ECE provision found elsewhere (see Morrissey et al., 2022), and may have implications for policies, such as targeting funding to specific geographic areas. It may be that in some communities public ECE programs narrow the relatively wide disparities in private ECE program access. Yet, importantly, HS enrollment tended to be lower in urban and suburban counties with large shares of Black children, and EHS enrollment was lower in counties with a larger share of Black children regardless of county urbanicity. These findings highlight the need to attend to racial equity in program expansions.

We also find evidence that public programming tends to be more stable than private. When public program availability is volatile, it tends to be due to large expansions rather than contractions. The relative stability of public ECE programs compared to private may serve a role in promoting equity, given that public programs are generally targeted to children in poor or low-income households. However, more research is needed regarding how program stability affects individual children.

Generally, we find that more disadvantaged communities have less access to ECE programs or services than more advantaged communities (Malik et al., 2018; NSECE Research Team, 2016). High-poverty counties have less preschool participation and less private ECE provision compared to low-poverty counties; public programs are somewhat effective in addressing these gaps but disparities across county poverty levels remain. Enrollment in public pre-K is relatively similar across counties with low and high child poverty rates, but lower in those with a moderate child poverty rate (between 10 and 20%). Counties with high child poverty rates average significantly more slots in HS centers. These patterns may reflect the targeting of HS and public pre-K to poor children, but local school district capacity or willingness to operate public pre-K programs likely plays a role as well. Surprisingly, EHS slots did not follow this pattern, but were relatively higher in counties with lower child poverty rates. This may reflect the severe lack of infant and toddler ECE services across all counties (e.g., Chaudry et al., 2021). Together, these results offer suggestive evidence that Head Start enrollment counteracts disparities in private ECE provision between high- and low-poverty counties. EHS does not appear to counteract these disparities to the same degree.

Our findings around ECE provision and the racial or ethnic composition of communities are nuanced. We find that counties where Black children comprise a larger share of the population have higher preschool enrollment rates. It is possible that the higher rate of participation in the child care subsidy among Black children (Chien, 2024) may at least partially underlie this higher enrollment in preschool, though our analysis does not directly speak to this causal conclusion. Regardless of racial or ethnic composition, there is consistent evidence of a stronger nonprofit ECE sector, and generally higher private preschool enrollment, in urban counties than in suburban rural counties. We find evidence, however, that public pre-K, HS, and EHS per capita enrollments are highest in rural counties regardless of racial and ethnic composition. Differences in public pre-K enrollment across county racial and ethnic composition varied across urban, suburban, and rural counties and did not exhibit a consistent pattern. We find relatively consistent evidence that urban and suburban counties where Black and Hispanic children composed a larger share of the population had fewer HS and EHS slots per capita; these results were more mixed in rural counties. These findings should all be interpreted with caution for two important reasons. First, our county-level analyses may obscure inequalities in ECE access that result from racial and ethnic segregation at the neighborhood level. Second, we do not capture quality of ECE programs, a critical dimension of ECE that has also been found to vary spatially in ways that disadvantage children of color (McCormick et al., 2023).

Combined, our results illustrate how the complex localized landscape of ECE provision, composed of public and private programs with different funding structures and operational realities, can produce disparities in ECE access. Given the evidence that participation in high-quality ECE supports parental employment and children’s educational, health, and economic outcomes, these disparities have numerous immediate and downstream implications for equity (Chaudry et al., 2021; Phillips et al., 2017; Yoshikawa et al., 2013). Evidence of spatial variation in ECE programming mirrors household-level research, which finds participation in preschool and center-based child care remains lower among children from lower-income families compared to their higher-income peers, with implications for school readiness and later outcomes (Chaudry et al., 2021; Yoshikawa et al., 2013). Findings reported here join a growing body of research that strongly suggest the need for federal and state policymakers to address the geographic inequities in the funding and provision of early childhood programming. Specifically, current federal funding formulas for programs such as Head Start and Early Head Start should be revised to provide additional funding to areas experiencing increases in poverty or demand for subsidized care (U.S. Government Accountability Office, 2024).

We also believe our findings have implications for ongoing early childhood policy research. Future research should extend beyond spatial variation in EHS and HS capacity (i.e., slots per poor child), to investigate geographic variation in program type, quality, and services. For example, PIR data contain information on the availability of full-day services, important for parental employment, and teacher turnover, an indicator of ECE quality. Such work would advance inquiry into racial disparities in program quality (Friedman-Krauss et al., 2016; McCormick et al., 2023). And, as noted earlier, there are few high-quality, longitudinal data sources for ECE availability, costs, and enrollment, reflecting the fragmented ECE system. Improved collection and dissemination of high-quality data is imperative to better understand access to ECE and inequalities in ECE provision. For example, while our survey data capture self-reported enrollment in private ECE settings for 3- and 4-year-olds, we are not able to capture most private for-profit or home-based ECE settings through administrative data. ACS survey data also lacks information about ECE participation among infants and toddlers under 3. Of additional value would be individual-level microdata about ECE attendance or information at more granular levels of geography. For example, more local analyses, particularly those using microdata to understand individual families’ proximity to ECE services, could explore the effects of neighborhood segregation by race and ethnicity on equity in ECE access. More broadly, these research advances could all be used to develop a greater understanding of racial and ethnic inequalities in ECE access. Further, our data are from the pre-COVID-19 period; analyses with more recent data could shed light on the myriad of changes in the ECE market in the wake of the pandemic and recovery. Building infrastructure to collect and publish high-quality comprehensive, longitudinal, spatially-linked data on ECE would support future research efforts, including linking program availability, participation, and child outcomes.

6. Conclusion

In sum, this study offers a descriptive examination of how participation, enrollment, and funding of early childhood programs varies over local geography. Families living in rural counties, and to some degree those in higher-poverty counties, face particularly reduced access to private and nonprofit ECE. Federally funded programs like EHS and HS serve a particularly important role in providing access to early childhood education in these communities, but these programs serve only a fraction of those eligible.

Improvements to ECE data infrastructure and reporting would enhance knowledge about ECE inequities and how to best target resources, as well as build understanding of how variation in provision relates to children’s outcomes. In addition, as more data emerges from the pandemic and post-pandemic period, it will be critical to examine how the COVID-19 crisis and public pandemic relief affected ECE programs and their geographic availability.

Supplementary Material

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.ecresq.2024.09.006.

Data availability

Data will be made available on request.