Introduction

Access to affordable childcare services plays a pivotal role in influencing both workforce participation and the broader economic outlook, especially for working parents. In the United States, the absence of national-level resources compels parents to independently manage and finance daytime care for their children, significantly impacting their capacity to work and save for the future. Our research aims to examine how the affordability of childcare aligns with a complex web of socioeconomic factors, including job sector, poverty rate, parental demographics, and geographic location. This research not only sheds light on the economic implications for families but also serves as a valuable resource for policymakers seeking to build supportive and sustainable childcare systems across the nation.

Data

The dataset we use comes from the National Database of Childcare Prices, compiled by strategic consulting company ICF. It includes county- and state-level demographic and labor data from the American Community Survey and state childcare market price surveys, which is where the childcare price information comes from.

There are 34567 observations in the dataset, each of which corresponds to study results from a county in the U.S. at a given year between 2008 and 2018 inclusive. The particular variables we use from this dataset are:

  • State_Name: Name of the U.S. state

  • County_Name: Name of the U.S. county

  • Study_Year: Year in which the data for the observation was collected

  • UNR_20to64: Unemployment rate of the population aged 20 to 64 years old.

  • EMP_M: Percent of civilians employed in management, business, science, and arts occupations aged 16 years old or older in the county.

  • MCInfant: Aggregated weekly, full-time median price charged for Center-based Care for infants (i.e. aged 0 through 23 months)

  • MCToddler: Aggregated weekly, full-time median price charged for Center-based Care for toddlers (i.e. aged 24 through 35 months)

  • MCPreschool: Aggregated weekly, full-time median price charged for Center-based Care for preschoolers (i.e. aged 36 through 54 months)

  • MCSA: Weekly, full-time median price charged for Center-Based Care for those who are school age based on the results reported in the market rate survey report for the county or the rate zone/cluster to which the county is assigned

  • MHI: Median household income

  • PR_F: Poverty rate for families

  • H_Under6_BothWork: Number of households with children under 6 years old with two parents that are both working

  • H_Under6_SingleM: Number of households with children under 6 years old with two parents with only the father working

  • H_Under6_FWork: Number of households with children under 6 years old with two parents with only the father working

  • H_Under6_MWork: Number of households with children under 6 years old with two parents with only the mother working

  • Households: Number of households in the county

  • FME: Median earnings for females for the population aged 16 years old or older

Research Questions

Our report aims to explore the following three questions:

QUESTION 1 How does employment in professional-managerial occupations influence income across different regions in the U.S, and what are the mitigating effects of unemployment rates on this relationship?

QUESTION 2 Which types of households will struggle to pay higher prices for childcare based on their income and household structure?

QUESTION 3 How do socioeconomic factors influence childcare costs across U.S. states, and what are the implications for accessibility and affordability for disadvantaged families?

Analysis

We examine each of the three questions in detail through a diverse set of statistical visualizations and analyses.

Question 1

This graph examines the distribution of Median Household Income across four groups of counties differentiated by PMC (people in management, business, science, and arts occupations) levels and unemployment rates: high PMC/high unemployment, high PMC/low unemployment, low PMC/high unemployment, and low PMC/low unemployment. It’s evident, though not conclusively, that counties with high PMC and low unemployment generally have higher median household incomes. In contrast, counties with high PMC and high unemployment show a wider spread in income, including lower median incomes but with some counties still achieving high incomes. Counties with low PMC, regardless of their unemployment status, tend to have less variation in income distribution. Notably, counties with low PMC and high unemployment have the lowest median household incomes, suggesting that a higher PMC presence might buffer against the economic effects of high unemployment.

The choropleth map provides a bivariate analysis exploring the correlation between the proportion of the professional-managerial class (PMC) and median household income across different states. The visualization classifies states into four categories based on whether they fall above or below the median values for PMC employment and median household income.

The predominant observation is a clear contrast: states are mostly classified as either having both high PMC employment and median household income or low in both categories. This pattern underpins the theory that states with a higher concentration of professional-managerial occupations tend to exhibit elevated household incomes, supporting the narrative that knowledge-based economies potentially yield higher earnings.

There appear to be some interesting regional trends. Most low PMC and low median household income states are in the South except for Michigan, Arizona, and Idaho, with every Southern state except for Virginia and Florida having a low PMC percentage and low median household income. Most of the high median household income but low PMC states are in the Rust Belt except for Nevada. Most of the high median household income and high PMC states are either in the Northeast, the Plains states, or the West Coast (except for Oregon), with Virginia (probably because of its affluent beltway population), Utah, Colorado, and Wyoming also having high PMC and high median household income.

However, there are intriguing exceptions that deviate from this trend. For instance, states like Florida and Washington, despite boasting higher PMC percentages, have median household incomes below the national average. This could be influenced by regional educational disparities, industry saturation, or an overrepresentation of dual income, no kids households inflating median income statistics. Additionally, former Rust Belt states like Ohio, Indiana, and Illinois, with traditionally strong unions, maintain high median household incomes despite lower PMC percentages, suggesting that well-paying blue-collar jobs still play a significant role in these economies.

In our examination of the association between PMC employment and household income across the U.S., we applied analysis including histograms and choropleth maps. The results indicated that regions with higher PMC employment and lower unemployment rates generally enjoy higher median household incomes, pointing to the economic advantages of professional-managerial jobs. Conversely, areas with low PMC presence, particularly coupled with high unemployment, face the most significant economic challenges, as reflected in the lower median household incomes. This suggests a correlation where a robust PMC sector may mitigate the adverse economic effects of unemployment. These insights highlight the nuanced impacts of job class on economic prosperity and could inform targeted economic development strategies.

Question 2

We aim to answer the question: which types of households will struggle to pay higher prices for childcare based on their income and household structure? Specifically, we examine different factors that might affect single mothers or households where only the mother works and their ability to pay for childcare prices.

We examined county’s median household income compared to proportion of households in a county with one or more children under 6 years old. We find that overall, it seems to be more common for households to have both parents to be working, slightly less common to have only a father working, slightly less common to have a single mother working, and least common to have only a mother working. This lines up with our understanding, as it is more common to have a single mother only supporting children than a mother supporting another parent and their children, as historically when one parent works it is usually the father working.

We decided to assess how many households there are of each type of parental situation. We again see that households with both parents working are the most common. We also see that the number of households with only fathers working has overall decreased, while households with both the parents working has increased in the most recent years from 2016 to 2018, and households with only the mother working has increased just barely. Explanations for this could be that prices of living have risen, requiring both parents to work, or that we as a society have moved away slightly from more patriarchal beliefs or households, leading to an increase in women working overall (contributing to the rise of both parents working and only mothers working).

We then investigate the female median earnings in each county and compare it with the number of households in that county consisting of single mothers working or only a mother working with presumably a spouse to take care of children. We adjusted the data on a logarithmic scale in order to better examine the differences across large numbers. We find that overall, counties with more single mothers working seem to have slightly lower female median earnings compared to counties with more mothers working. This is a bad situation, since single mothers don’t have a spouse that can help take care of the children, and therefore are likely in greater need of childcare.

Overall, we find that there are relatively high numbers of single mothers, especially when compared to households with only mothers working, that may be earning less than mothers with spouses, and that are likely to be in greater need of childcare.

Question 3

In order to explore the third question, we will examine the variables of median price of center-based care for infants, poverty rate for families, and median household income across states in the contiguous United States.

We fit smoothed density curves using the loess method to plot the median weekly price of center-based care for infants against the poverty rate of families for all age groups, ranging from infants to school-age children. We have also indicated where the average poverty rate, which is around 11.7%, falls relative to the curves.

From the plot above, we can see that there does appear to be a relationship between poverty rate and median price for centers, and this relationship varies based on the age group being cared for. In general, prices are highest for states where the poverty rate is very low, due to families being able to pay more. Prices drop as the poverty rate increases, but start to pick up again when the poverty rate is around 40%. It also appears to be the case that younger children like infants and toddlers are more expensive to care for in wealthier states, but the gap between age groups narrows as poverty rate goes up. It can be inferred that poor families with young children in wealthier states will have a much harder time paying for center-based care.

We created a choropleth map showing the burden of childcare on family income at the state level. This variable was calculated by multiplying the median weekly price of center-based care for infants by 52 to get the yearly price, and then dividing by the median household income for each state. We also scaled it to be a percentage.

It is apparent that the burden of childcare is especially large for coastal states like Washington, California, and Massachusetts, where families would need to spend between 24% and 28% of their household income in order to have their infant be cared for at a center. With the exception of Colorado, most of the states in the Great Plains have a burden below the median, with South Dakota and Kansas being especially low in the 10% to 12% range. In general, the East Coast states hover slightly above the median, while some states like Florida, Oklahoma, and Montana have a median burden.

Poorer families living in expensive states like California and Washington likely struggle to find affordable childcare that does not consume a significant portion of their income, possibly driving them to use childcare services that are of a lower quality and safety. These states, however, have the advantage of having better job opportunities as a draw, though it is dampened by the exorbitant burden of childcare.

In order to assess the significance of the observed differences in the choropleth map, we conducted a formal statistical analysis in the form of a visual randomization test. The null hypothesis states that any differences are due to chance, while the alternative hypothesis states that they are the result of an underlying pattern in the data. We repeatedly generated choropleths based on a random sample of the data and randomly plotted them in a grid next with the original map in the mix.

Upon visual comparison (tested by third parties) of the resulting grid above, the real map does stand out from the rest as being the one on the top left. This is based on the fact that it is the only state that maintains coherent clusters of low percentages on the Great Plains and higher percentages on the Coasts, which make sense. Most of the other maps also have randomly placed states with median percentages relative to their neighbors. This allows us to conclude that there is a meaningful relationship between geographic location and/or state policies and the burden placed by childcare on a family.

Given what we have observed about the effects of poverty rate and childcare burden, we produced a scatterplot with a linear curve plotting the infant center-based childcare burden against family poverty rate. The general trend appears to be that a higher poverty rate is indeed associated with a larger burden placed on families. There is some variance in the data towards larger burdens for both low and high poverty rate, and some variance below the line for poverty rates around 40%, similar to the trend observed in the first plot where childcare prices begin to drop when poverty increases to a near-majority.

In our analysis of the third question, we utilized visualizations such as density curves, choropleth maps, and scatterplots in order to explore the relationship between childcare costs, poverty rates, and median household income across U.S. states. The findings suggested that younger children in wealthier states had higher care costs, indicating that poorer families in expensive states may struggle to afford childcare without compromising quality. As poverty rates increase, childcare prices decrease, but there is still a positive association between family poverty rates and childcare burdens, indicating that higher poverty rates are linked to larger burdens on families.

Conclusion

Results

We began our report by conducting an analysis of PMC employment and household income across the U.S, which revealed that regions with higher PMC employment and lower unemployment rates tend to have higher median household incomes, suggesting potential economic benefits of a robust professional-managerial sector. Next, we examined the effects of household structure on income, and the potential implications for childcare requirements. We found that there is a large number of single working mothers who are in greater need of childcare but also tend to earn less than working mothers with stay at home spouses. Finally, we found that childcare costs are higher for younger children in wealthier states, potentially posing challenges for poorer families located there. While childcare prices decrease with increasing poverty rates, there remains a positive association between family poverty rates and childcare burdens, indicating larger financial challenges for families with higher poverty rates.

Overall, these findings suggest a greater need for resources dedicated to the support of single mothers working to support their children and families with an income below the poverty level. Furthermore, the geographic and regional trends we observed lead us to conclude that these resources should be managed and directed on a state level, in order to best meet the needs of the local population.

Limitations and Further Research

Our study comes with several limitations. Some limitations are from the dataset collection itself: the group that compiled the dataset we used made several standardizations and imputations due to the differences in how data was collected or how demographics were defined in different states and counties. Variables that include age groups and market prices are likely to be affected. Some limitations are from what data was available to us: we were unable to partition childcare price data by demographics like race of population, labor sectors of population, and gender of parents working in households with infants, which heavily limited our analysis.

In the future, we hope to collect more data on childcare prices by demographics like race, labor sector, and household structure. This would enable us to better pinpoint which groups need the most assistance with high childcare prices, focusing advocacy efforts and conserving resources. Additionally, since the data we use only extends from 2008-2018, we would hope to collect more recent data so we could conduct a more comprehensive time analysis.