Racial and Ethnic Discrimination in the Labor Market for Child Care Teachers

Abstract

This article examines racial and ethnic discrimination in the child care teacher hiring process. We construct a unique data set that combines a résumé audit study of center-based providers with a follow-up survey of those in the original audit sample. Fictitious résumés were randomly assigned White-, Black-, and Hispanic-sounding names and submitted in response to real teacher job advertisements. The survey was then administered to capture the characteristics of children, teachers, and administrators within the center. These data reveal three key results. First, we find robust evidence of discrimination: Black and Hispanic applicants receive significantly fewer interview requests than observationally equivalent Whites. Second, our results are consistent with a model of customer discrimination: The racial and ethnic composition of the center’s customer base is correlated with the characteristics of job seekers receiving an interview. Finally, we show that states’ child care regulations mitigate the racial and ethnic gap in interview requests.

Keywords

center-based child care discrimination in hiring diversity early childhood economics of education ethnicity experimental research race racial discrimination résumé audit study survey research teacher characteristics

This article provides a comprehensive analysis of racial and ethnic discrimination in the hiring of center-based child care teachers. Specifically, we implement a unique two-step data collection procedure that combines a résumé audit study of child care centers in 14 U.S. cities with a detailed follow-up survey of providers in the original audit sample. Fictitious résumés were randomly assigned predominately White-, Black-, and Hispanic-sounding names that were submitted in response to real teacher job advertisements. The provider survey was then administered to inquire about the characteristics of children, teachers, and administrators within the center. Together, these data allow us to study three features of discrimination within the child care labor market. We begin by examining whether providers engage in discrimination when hiring teachers and the mediating influence of program directors’ race and ethnicity on hiring decisions. Second, given the nature and intensity of interaction between child care staff and parents, we provide an empirical test of the role of customer discrimination in teacher hiring. Our final set of analyses examines the impact of states’ child care regulations on racial and ethnic disparities in interview requests.

Although the résumé audit methodology has been used extensively to examine racial and ethnic discrimination across a variety of labor markets and countries (e.g., Bartoš, Bauer, Chytilová, & Matějka, 2016; Bertrand & Mullainathan, 2004; Carlsson & Rooth, 2007; Nunley, Pugh, Romero, & Seals, 2014; Oreopoulos, 2011), we believe for several reasons that such an investigation is particularly important in the child care market.

First, the emergence of an achievement gap between minority and nonminority children during preschool-age years has increased the urgency of adopting early childhood policies that deliver high-quality, culturally competent programming (Fryer & Levitt, 2013). One proposal gaining traction is to increase the hiring and retention of minority child care teachers. Researchers and administrators posit that doing so will provide developmental benefits for minority children by exposing them to demographically similar “role model” adults who may serve as mentors and advocates. In addition, minority teachers may be less likely to exhibit unconscious forms of bias that can impede productive child–teacher interactions, and they may serve as stabilizing cultural “bridges” between children’s home and child care environments.

Consistent with these hypotheses, there is mounting empirical evidence that minority children particularly benefit when they are matched with teachers from the same racial or ethnic background. Although the literature focuses primarily on grades K–12, evidence suggests that such race matching generates short-run improvements in test scores (Dee, 2004; Egalite, Kisida, & Winters, 2015), classroom behavior (Wright, Gottfried, & Le, 2017), attendance and suspensions (Holt & Gershenson, 2015; Lindsay & Hart, 2017), teacher evaluations of student performance and expectations (Dee, 2005; Gershenson, Holt, & Papageorge, 2016), and academic attitudes (Egalite & Kisida, 2018). Furthermore, Gershenson, Hart, Constance, and Papageorge (2017) show that race matching in primary school may generate long-run benefits by reducing high school dropout rates, particularly among Black males.

The smaller literature on preschool programs also finds that race matching is important for young children’s educational experiences. Two studies of prekindergarten programs show that race matching is positively associated with teacher reports of children’s initial program readiness (Downer, Goble, Myers, & Pianta, 2016) as well as short-run gains in academic and social-emotional skills (Downer et al., 2016; Graves & Howes, 2011). Another study finds that children matched to same-race teachers are rated to have more positive interactions—and closer overall relationships—with their teachers (Saft & Pianta, 2001). A related stream of work finds that minority teachers hold higher expectations for and are more optimistic about the academic futures of children from their own racial and ethnic group (Gilliam, Maupin, Reyes, Accavitti, & Shic, 2016; Murray, Murray, & Waas, 2008; Tenenbaum & Ruck, 2007).

The salience of these issues—the early emergence of racial achievement gaps and teacher-based policies to ameliorate them—will continue to grow in the coming decades given the increasing racial and ethnic diversity of the preschool-age population.¹ Yet despite this growing diversity, the non-White share of the early education workforce continues to lag behind that of the children enrolled in these programs.² To the extent that discrimination exists in the child care labor market, it may prevent the workforce from attaining the level of diversity needed to serve the developmental needs of an increasingly heterogeneous population. Thus our first goal in the article is to document whether there are racial and ethnic disparities in child care teacher hiring. In addition, we examine to what extent the race and ethnicity of the center director influences the relative treatment of minority applicants. Given that an overwhelming majority of program directors are White, it is plausible that own-race preferences provide an explanation for the differential treatment of minorities.³

The second motivation for studying discrimination in child care hiring is that the production of early education services is highly labor-intensive. This trait is appealing for our purposes because there is a high degree of face-to-face interaction between employees (i.e., teachers) and consumers (i.e., parents and children). Furthermore, communication between teachers and parents likely relates to sensitive topics about children’s health and development. Such dynamics create the potential for consumers’ racial preferences to shape hiring behavior of child care providers. Indeed, Becker’s (1957) seminal work on discrimination posits that some consumers have discriminatory tastes—preferring to receive a given service from members of their own racial and ethnic group—which in turn may lower the demand for workers from other groups, reduce their relative wages, and lower the market price charged by firms that continue to hire minority individuals. If these conditions are present in the child care market, profit-maximizing firms have two options: hire minority teachers at low wages and charge lower rates to nondiscriminatory parents or hire high-wage nonminority teachers and charge higher rates to discriminatory parents.

There are other reasons to believe that parents’ child care decisions are shaped by racial preferences. First, although parents claim to value high-quality, education-based child care programs, the evidence suggests that actual consumption decisions are driven by a range of nonquality factors, including parents’ own personal biases and cultural beliefs (Caldera & Hart, 2004; Chaudry et al., 2011; Pungello & Kurtz-Costes, 1999; Mamedova & Redford, 2015).⁴ In addition, parents generally allocate little time to the child care search, considering just one or two providers before making a decision (National Survey of Early Care and Education Project Team, 2014). Together, these dynamics suggest that parent decisions may be influenced by such easy-to-observe features of the child care environment as staff racial and ethnic composition. If parents’ discriminatory tastes lead to differential sorting based on teachers’ observable characteristics, then providers have a strong incentive to adjust hiring behavior in a way that accords with consumer preferences. Therefore, another goal of the current article is to understand whether customer discrimination plays a role in hiring decisions. In particular, we examine to what extent the racial and ethnic composition of children attending the center is correlated with the gap in interview requests between minority and nonminority applicants.

The final motivation for studying this topic is that the child care market is characterized by high levels of regulation. All states set minimum standards for child-to-staff ratios, maximum group sizes, and staff qualifications. Such forms of occupational licensing are aimed at mitigating well-known information asymmetries in the child care market, in which parents are ill informed about the negative external benefits generated by low-quality care as well as how to identify the attributes of high-quality providers (Bassok, Markowitz, Player, & Zagardo, 2018; Mocan, 2007). In markets with imperfect information, sellers have an incentive to produce lower-quality services by hiring less productive workers and underinvesting in their skill acquisition (Blau, 2001; Shapiro, 1986). As a result, sellers may be more inclined to indulge their distaste for hiring minority applicants than would be the case if consumers were perfectly informed. By reducing consumer search frictions, occupational licensing increases the cost of discriminating against well-qualified members of minority groups. This discussion suggests that various forms of licensing may have the unintended consequence of reducing racial and ethnic disparities in the child care labor market, a proposition that finds empirical support in other labor markets (e.g., Blair & Chung, 2017).

Therefore, the final goal of this article is to test whether child care regulations influence the hiring behavior of center-based providers. Although there is a sizable literature studying the impact of regulations on the child care market (e.g., Blau 2001, 2003; Hotz & Kilburn, 1994; Hotz & Xiao, 2011), ours is the first study to consider their influence on race-based hiring preferences. Specifically, we focus on regulations governing the labor-intensiveness of child care provision (i.e., child-to-staff ratios and maximum group sizes) and staff qualifications (i.e., experience and education requirements). These domains were chosen because they are likely to be given significant consideration by directors during the hiring process. A key goal of our analysis is to test for heterogeneous effects of regulations across low- and high-quality applicants as well as across providers located in low- and high-income markets. Indeed, as will be discussed, it is possible that extremely well-qualified minority applicants, particularly those applying in high-income communities, may benefit the most from tighter regulations.

Data Sources

Résumé Audit Study

The current article is part of larger project that administered a résumé audit study to understand teacher hiring practices in the market for center-based child care (Boyd-Swan & Herbst, 2018). The setting for our audit study is a large online job board in the United States. We used this website to search for child care teacher job advertisements in 14 large cities: Atlanta, Boston, Chicago, Dallas, District of Columbia, Houston, Los Angeles, Minneapolis, New York City, Philadelphia, Phoenix, San Diego, San Francisco, and Seattle. Fieldwork for the audit study began in May 2016 and ended in January 2017. Our goal was to submit four résumés in response to each job advertisement. Altogether we submitted 10,986 résumés in response to 2,772 job advertisements, of which 2,720 (98.1%) received all four résumés.

We submitted résumés in response to postings for early childhood education (ECE), child care or daycare lead teachers, assistant teachers and aides, co-teachers, and floating-classroom teachers. These positions were located in infant, toddler, or preschool-age classrooms as well as before- and after-school settings. In addition, we limited the job search to child care taking place in for-profit and nonprofit centers, places of worship, community-based organizations, and school-based before- and after-school programs.

Using software created by Lahey and Beasley (2009), we generated a large number of fictitious résumés, each one containing five sections of randomly assigned characteristics.⁵ For the purposes of this study, the key résumé attribute is the assignment of a predominately White-, Black-, or Hispanic-sounding name. Given that about 95% of child care teachers are women, we examine only female-sounding names (Boyd-Swan & Herbst, 2017a). Data on the most common surnames by race and ethnicity were collected from the 2000 U.S. decennial census.⁶ Given that the most common forenames by race and ethnicity are not reported by the U.S. Census Bureau, we relied on the New York City Health Department’s Bureau of Vital Statistics birth records from 2014, Babycenter.com’s 2016 list of popular names, and the list used by Bertrand and Mullainathan (2004).⁷ Names within each racial and ethnic category were randomly assigned with equal probability (1/3), and individual names were drawn without replacement, thereby ensuring that no duplicates appear in a set of four résumés.⁸

It is widely acknowledged within the audit study literature that résumé names may signal not only the race and ethnicity of a given applicant but also his or her socioeconomic status (SES). Therefore, estimates of the effect of a given racial category may confound employer preferences for applicants’ race with background characteristics. Indeed, Bertrand and Mullainathan (2004) provide evidence of substantial name-based heterogeneity in SES: Highly educated mothers tend to choose different names compared to their less educated counterparts. However, the authors find no relationship between the implied SES of applicant names and the likelihood of receiving an interview. Thus the authors conclude that traits such as SES do not explain the White–Black gap in interviews. Nevertheless, we take some steps to minimize the possibility that child care providers select applicants on the basis of perceived SES. All mailing addresses within a city reside in a zip code that is at or close to the median household income for the city as a whole. Therefore, all of our fictitious applicants within a city reside in the same medium-income area. We purchased actual addresses from a commercial vendor that maintains an extensive, up-to-date database of residential mailing addresses used primarily for targeted advertising campaigns. Each address was assigned with equal probability without replacement.

To allow child care providers to contact our fictitious job seekers, we established an e-mail account for each name used in the study. Research assistants monitored these accounts and recorded whether résumés received an e-mail message from a provider. Research assistants were trained to code interview requests, which for our purposes were defined as having received an explicit interview request or an invitation to discuss the résumé or position in more detail. In addition to coding employer responses, the research assistants recorded a variety of information about the job advertisements, including the minimum experience and education requirements for the position.

Child Care Provider Survey

Following the completion of the résumé field experiment, we administered a web-based follow-up survey to all of the providers in the original audit study sample. This survey was nominally independent of the audit study, in that providers were not made aware that they had participated in a different study. To our knowledge, this is the first work to implement such a two-step data collection procedure for the purpose of studying labor market discrimination.⁹

Research assistants used multiple online sources to search for and record the name of and contact information for the director of each child care center. In some cases, only the name of an assistant director (or other program administrator) or general contact information could be located; this information was used for survey distribution purposes. In a small number of cases (132), no contact information for the center or its personnel could be located; such centers were excluded from the survey. Data collection for the survey occurred between January and June of 2017. Each provider received a presurvey e-mail message, an initial request to complete the survey along with the link, and three reminder e-mails. Attempts at data collection were ceased after the third reminder. Full methodological details on the survey’s implementation can be found in Boyd-Swan and Herbst (2017a, 2017b).

The provider survey contained detailed questions about the characteristics of the center (e.g., auspice, enrollment, racial and ethnic composition of children, accreditations, and fees), the teaching staff (e.g., staff size, racial and ethnic composition, staff training and education, and compensation), and the center director (e.g., race and ethnicity, experience and education, and salary).¹⁰ As will be shown in later sections, such information was important for studying how the race and ethnicity of the director as well as the racial composition of actual consumers (i.e., children) influence teacher hiring behavior.

We received usable survey data from 514 child care providers (out of 2,340) for a response rate of 22%.¹¹ When these provider-level data are matched with the appropriate résumé-level data, we have a maximum of 2,047 résumés available for analysis. A key question is whether the providers completing the survey are equivalent to those that did not. To this end, we undertake two analyses. First, online Appendix Table 3 (available on the journal website) compares provider- and neighborhood-level characteristics across the survey responders and nonresponders. We find that those completing the survey are similar to their counterparts that did not. It is encouraging that providers’ minimum experience and education requirements—rough proxies for program quality—are quite comparable. In addition, the National Association for the Education of Young Children accreditation rate—a strong measure of quality—is virtually identical across the responders and nonresponders. Online Appendix Table 3 also reveals a close correspondence in programs’ neighborhood characteristics. Second, we examine the extent to which the randomly assigned résumé characteristics have similar effects on interview requests, as shown in online Appendix Table 4. Again, at least with regard to hiring preferences, both groups of providers seem comparable: Coefficients on the résumé characteristics are mostly consistent in sign and magnitude. Nevertheless, despite this evidence, some caution is warranted in interpreting our survey results. It is possible that providers not responding to the survey are different on unobservable dimensions (e.g., they may be more time or resource constrained), which could limit the representativeness of those included in the current study.

The Impact of Race and Ethnicity on Child Care Hiring Decisions

Applicant Race and Ethnicity and Interview Requests

This section establishes the causal effect of job-seeker race and ethnicity on interview requests in the market for child care teachers. Overall, 27% of résumés with White names were invited for an interview, compared to 20.1% for Black names and 23.8% for Hispanic names. This translates into a Black−White difference of −6.9 percentage points and a Hispanic−White difference of −3.2 percentage points. Both differences are statistically significant. Table 1 presents regression-adjusted estimates of the gaps in interview requests, derived from the following specifications:

\begin{matrix} {Interview}_{i k c m} = β_{0} + β_{1} {Black}_{i k c m} + β_{2} {Bispanic}_{i k c m} \\ + X' β + α_{c} + δ_{m} + λ_{k} + ε_{i k c m,} \end{matrix}

(1a)

\begin{matrix} {Interview}_{i k j} = β_{0} + β_{1} B l a c k_{i k j} + β_{2} H i s p a n i c_{i k j} \\ + X' β + λ_{k} + γ_{j} + ε_{i k j,} \end{matrix}

(1b)

where Interview is a binary indicator equal to 1 if résumé i submitted in response to job advertisement j in city c and month m received an explicit interview request and 0 otherwise. The variables Black and Hispanic are binary indicators for résumés with Black- and Hispanic-sounding names, respectively. The baseline model (Equation [1a]) includes a set of city fixed effects (α), month fixed effects (δ), and dummy variables for the order in which the ith résumé was submitted (λ). We also condition on the other randomly assigned résumé characteristics (X′). As a robustness check, we estimate a second model (Equation [1b]) that replaces the city and month fixed effects with job advertisement fixed effects (γ; FE model), which allows for the possibility that the résumé characteristics are correlated with provider characteristics. Thus the FE model uses only within-job-advertisement variation to generate the coefficients. All standard errors are clustered at the job advertisement level. Table 1 shows that applicants with Black- or Hispanic-sounding names are substantially less likely to receive an interview than their observationally equivalent counterparts with White-sounding names. Specifically, the full ordinary least squares (OLS) model in column (1) of Panel A shows that Black applicants are 7.7 percentage points less likely to receive an interview and Hispanic applicants are 3.3 percentage points less likely. The coefficients imply that, relative to the White applicant mean, the interview rate for Blacks is approximately 29% lower, and the interview rate for Hispanics is 12% lower. As expected, the FE model estimates in column (1) of Panel B are similar to those in the OLS model.

Table 1

Racial and Ethnic Differences in Interview Requests, by Teacher Type

	Full Sample	Assistant Teachers	Lead Infant/Toddler Teachers	Lead Preschool Teachers
Variable	(1)	(2)	(3)	(4)
Panel A: OLS estimates
Black name	−.0767*** (.0091)	−.1139*** (.0184)	−.0691*** (.0161)	−.0872*** (.0145)
Hispanic name	−.0333*** (.0094)	−.0401** (.0193)	−.0171 (.0164)	−.0400*** (.0149)
Panel B: Job advertisement FE estimates
Black name	−.0739*** (.0084)	−.0978*** (.0178)	−.0672*** (.0150)	−.0942*** (.0131)
Hispanic name	−.0273*** (.0085)	−.0296 (.0184)	−.0097 (.0146)	−.0347*** (.0133)
Observations	10,986	2,712	3,653	4,324

Note. Standard errors (in parentheses) are clustered at the job advertisement level. The omitted category includes résumés with White-sounding names. The models in Panel A include the other résumé characteristics, city and month fixed effects, and résumé order indicators. The models in Panel B omit the city and month fixed effects and include the job advertisement fixed effects.

p < .10. **p < .05. ***p < .01.

As previously mentioned, we submitted résumés in response to job advertisements for assistant and lead teacher positions. This is important given the differential skill requirements for individuals in these jobs. For example, 24% of assistant teacher positions required (at a minimum) an associate’s degree, and 6% required a bachelor’s degree. The comparable figures for lead teachers were 35% and 21%, respectively. Thus this study provides an opportunity to examine the racial and ethnic gap in interview rates across lower- and higher-skilled positions within the same occupation. Indeed, the most common practice in the discrimination literature is to examine skill-based differences in the interview gap across different occupations and industries (e.g., Bertrand & Mullainathan, 2004; Carlsson & Rooth, 2007).

Table 1 presents additional estimates on Black- and Hispanic-sounding names for the subset of assistant, lead infant/toddler, and lead preschool teacher advertisements, as shown in columns (2), (3), and (4), respectively. The lead teacher positions are further disaggregated by age group given the clear skill differences required for these jobs.¹² We find that the racial and ethnic interview gaps persist across these lower- and higher-skilled teacher positions. For example, estimates in Panel A imply a Black–White gap of –11.4 percentage points and a Hispanic–White gap of –4.0 percentage points in the market for assistant teachers (the lowest-skilled position), and the comparable gaps for lead preschool teachers (the highest-skilled position) are −8.7 and −4.0 percentage points, respectively.

The Influence of Program Directors’ Race and Ethnicity

Results from the preceding analyses consistently point to a racial and ethnic gap in interview requests. We now examine the impact of center directors’ race and ethnicity on interview requests; that is, we seek to understand whether the race and ethnicity of program directors influences this interview gap. Evidence from nationally representative surveys suggests that directors are overwhelmingly White, a finding that accords with our follow-up survey of the centers in the audit study. Approximately three-quarters of directors are White, 9% are Black, 9% are Hispanic, and the remaining 8% are other races and ethnicities. If center directors exhibit own-race biases during the hiring process, it is plausible that the interview gap between White and non-White applicants is explained by the significant presence of White directors within the center-based sector.

To examine this issue, we use information on directors’ race and ethnicity from the provider survey, matched to our résumé audit data, to estimate versions of the following regression model:

\begin{array}{l} {Interview}_{i k c m} = β_{0} + β_{1} White_director \\ + β_{2} (Black \times White_director) \\ + β_{3} (Hispanic \times White_director) \\ + β_{4} Black_director \\ + β_{5} (Black \times Black_director) \\ + β_{6} (Hispanic \times Black_director) \\ + β_{7} Hispanic_director \\ + β_{8} (Black \times Hispanic_director) \\ + β_{9} (Hispanic \times Hispanic_director) \\ + β_{10} other_director \\ + β_{11} (Black \times other_director) \\ + β_{12} (Hispanic \times other_director) \\ + α_{c} + δ_{m} + λ_{k} + ε_{i k c m,} \end{array}

(2)

where Interview, Black, and Hispanic are defined in the same manner as before, and variables White_director, Black_director, Hispanic_director, and other_director are indicator variables for White, Black, Hispanic, and other-race-ethnicity center directors, respectively. We estimate the model by interacting the indicators for director race-ethnicity with Black and Hispanic and including these interactions along with the director race-ethnicity controls in the regression. Omitted from the model are the main effects on Black and Hispanic so that the coefficient on the interactions can be interpreted as the percentage-point change in the likelihood that a Black or Hispanic applicant receives an interview (relative to a White applicant) when the center director is of a given race-ethnicity. The key motivation for excluding the main effects is that we are interested in understanding to what extent the race and ethnicity of the center director influences the likelihood that a Black or Hispanic applicant receives an interview, relative to a White applicant. In other words, we would like to know whether this provider-level contextual variable influences the Black–White and Hispanic–White interview gaps found in the previous analyses. Such a modeling approach provides an answer to this question. Given that Equation (2) controls for directors’ race-ethnicity, which is not randomly assigned, we present additional results from a model that replaces the city and month fixed effects with job advertisement fixed effects, in which only within-center variation in the interaction terms is used to generate the coefficients.

Estimates from the OLS model (column [1]) and FE model (column [2]) are presented in Table 2. Both sets of results are quite similar and provide evidence consistent with own-race bias in teacher hiring. As shown in the first set of results, White applicants are more likely than their Black counterparts to receive an interview when the center director is White (OLS, statistically significant 10.2 percentage points; FE, statistically significant 8.0 percentage points). However, White directors appear to be indifferent between White and Hispanic applicants (OLS, statistically insignificant 0.2 percentage points; FE, statistically insignificant 0.8 percentage points). Similarly, the next set of results show that Black directors favor Black over White applicants (OLS, statistically insignificant 14.8 percentage points; FE, statistically significant 20.4 percentage points), and they, too, are indifferent between White and Hispanic applicants (OLS, statistically insignificant −1.2 percentage points; FE, statistically insignificant 6.5 percentage points). Finally, when the center director is Hispanic, Hispanic applicants are more likely to receive an interview than their White counterparts (OLS, statistically significant 21.3 percentage points; FE, statistically insignificant 14.1 percentage points).

Table 2

Effects of Center Directors’ Race and Ethnicity on Interview Requests

Variable	(1)	(2)
White_director	.0521 (.0745)	—
Black × White_director	−.1017***^ (.0368)	−.0798**^ (.0322)
Hispanic × White_director	−.0022 (.0361)	−.0079 (.0308)
Black_director	−.0122 (.1067)	—
Black × Black_director	.1481 (.1219)	.2037* (.1106)
Hispanic × Black_director	−.0119 (.0948)	.0647 (.0904)
Hispanic_director	.1296 (.1168)	—
Black × Hispanic_director	−.0249^ (.1249)	−.0430^ (.1097)
Hispanic × Hispanic_director	.2126** (.1051)	.1411 (.1027)
Other_director	.0594 (.1130)	—
Black × Other_director	−.0102 (.0880)	−.0625 (.0842)
Hispanic × Other_director	−.0436 (.1173)	−.0657 (.1303)
Observations	1,217	1,217

Note. Standard errors (in parentheses) are clustered at the job advertisement level. Column (1) estimates the model using ordinary least squares (OLS), and column (2) includes job advertisement fixed effects. The OLS model includes the city and month fixed effects and résumé order indicators. The fixed-effects model omits the city and month fixed effects and includes the job advertisement fixed effects.

Indicates that a given pair of interaction coefficients (across Black and Hispanic applicants) is statistically significantly different at the .10 level or higher.

p < .10. **p < .05. ***p < .01.

Customer Discrimination

We now turn our attention to whether there is evidence of customer discrimination in the labor market for child care teachers. Two approaches have been used by previous studies to test for customer discrimination. The predominant methodology examines discrimination indirectly by looking at how the racial and ethnic composition of the potential pool of customers influences hiring decisions, where the customer pool is measured by local area race-ethnicity shares (e.g., Bertrand & Mullainathan, 2004; Carlsson & Rooth, 2007; Leonard, Levine, & Giuliano, 2010). A smaller set of studies takes a more direct approach by observing the characteristics of actual customers (Bar & Zussman, 2017; Holzer & Ihlanfeldt, 1998), and to our knowledge, only one previous study examines the composition of actual customers in the context of a field experiment (Neumark, Bank, & Van Nort, 1996).

The current study performs both analyses, thereby providing an opportunity to compare results derived from potential versus actual consumers of center-based child care programs. Specifically, we begin with the standard approach in the literature by analyzing whether the racial and ethnic composition of the local population (i.e., potential customers) is taken into account when program directors make hiring decisions. We then exploit the child care provider survey to examine whether the composition of children attending the center (i.e., actual customers) influences hiring behavior.

To implement the first approach, we geocoded the location of each child care center included in the résumé audit study and assigned to each one the appropriate census tract identification number. Using the 2010 U.S. decennial census, we calculated the fraction of each tract’s population that is White, Black, and Hispanic. We then estimated regressions of the following form:

\begin{array}{l} {Interview}_{i k c m} = β_{0} + β_{1} White_share \\ + β_{2} (Black \times White_share) \\ + β_{3} (Hispanic \times White_share) \\ + β_{4} Black_share + β_{5} (Black \times Black_share) \\ + β_{6} (Hispanic \times Black_share) \\ + β_{7} Hispanic_share \\ + β_{8} (Black \times Hispanic_share) \\ + β_{9} (Hispanic \times Hispanic_share) \\ + α_{c} + δ_{m} + λ_{k} + ε_{i k c m,} \end{array}

(3)

where Interview, Black, and Hispanic are defined in the same manner as before, and the variables White_share, Black_share, and Hispanic_share denote the fraction of Whites, Blacks, and Hispanics, respectively, within a given census tract. To estimate the model, we interact these race-ethnicity shares with the indicators for Black and Hispanic and include these interactions along with race-ethnicity shares in the regression. We suppress from the model the main effects on Black and Hispanic so that the coefficient on the interactions can be interpreted as the percentage-point change in the likelihood that a Black or Hispanic applicant receives an interview as the proportion of the census tract population of a given race-ethnicity increases from 0% to 100%. Equation (3) also includes a control for center director race and ethnicity in light of the results from the previous section showing the importance of such characteristics on teacher interview decisions. As an additional robustness check, we estimate a model that replaces the city and month fixed effects with job advertisement fixed effects.

Results from Equation (3) are presented in Table 3, which show the interactions on White_share, Black_share, and Hispanic_share within the census tract. Column (1) provides the baseline OLS estimates, column (2) adds controls for director race-ethnicity, and column (3) adds the job advertisement fixed effects. Once again, the estimates are similar across the specifications. The first cluster of coefficients, which shows the interactions of Black and Hispanic with White_share, provides evidence consistent with White customer discrimination: As the share of Whites within a census tract increases from 0% to 100%, minority applicants—particularly Blacks—become less likely than their White counterparts to receive an interview (Blacks, statistically significant 10.1 percentage points less likely; Hispanics, statistically significant 3.9 to 4.7 percentage points less likely). The second and third set of results, which shed light on Black and Hispanic customer discrimination, respectively, are less conclusive. Indeed, coefficients on the interactions are often small in magnitude, not precisely estimated, and not always positively signed. Thus a tentative conclusion is that program directors may account for the White customer base when making teacher hiring decisions—by favoring White applicants—but are less likely to consider the Black and Hispanic customer base—by not favoring Black and Hispanic applicants—when making these decisions. Alternatively, the results may also imply that customers of all races reveal a modest preference for White teachers.

Table 3

Tests of Customer Discrimination, by Census Tract-Level Measure of Race and Ethnicity Shares

Variable	(1)	(2)	(3)
White share of kids	−.0504 (.0543)	−.0450 (.0536)	—
Black × White_share	−0.1010***^ (.0166)	−0.1009***^ (.0166)	−0.1010***^ (.0152)
Hispanic × White_share	−.0478*** (.0173)	−.0468*** (.0173)	−.0388** (.0157)
Black share of kids	−0.1376** (.0576)	−0.1339** (.0569)	—
Black × Black_share	−.0639* (.0362)	−.0638* (.0363)	−.0170 (.0346)
Hispanic × Black_share	−.0346 (.0335)	−.0333 (.0334)	−.0110 (.0326)
Hispanic share of kids	−.0878 (.0662)	−.0905 (.0652)	—
Black × Hispanic_share	−.0275 (.0418)	−.0263 (.0419)	−.0637* (.0382)
Hispanic × Hispanic_share	.0048 (.0503)	.0025 (.0501)	−.0225 (.0469)
Observations	10,913	10,913	10,913
Director race-ethnicity	No	Yes	No

Note. Standard errors (in parentheses) are clustered at the job advertisement level. Columns (1) and (2) estimate the model using ordinary least squares (OLS), and column (3) includes job advertisement fixed effects. Column (2) adds a control for the race and ethnicity of the child care program director. The OLS models include the city and month fixed effects and résumé order indicators. The fixed-effects model omits the city and month fixed effects and includes the job advertisement fixed effects.

^Indicates that a given pair of interaction coefficients (across Black and Hispanic applicants) is statistically significantly different at the .10 level or higher.

p < .10. **p < .05. ***p < .01.

We now make use of the provider survey, which included questions about the total enrollment within centers as well as enrollments by race and ethnicity. We used this information to construct variables for the share of White (non-Hispanic), Black (non-Hispanic), and Hispanic children within the center. These variables were then interacted with Black and Hispanic in the same manner as earlier. On average, approximately 42% of children attending the centers are White, 14% are Black, and 9% are Hispanic.13 Results based on these variables are presented in online Appendix Table 5. Consistent with the census tract–level results, the first cluster of coefficients provides evidence in favor of White customer discrimination: As the share of White children within a center increases, White applicants are increasingly favored over their Black and Hispanic counterparts. The second set of results is also consistent with those in Table 3, in that the evidence does not point to the presence of Black customer discrimination. In fact, if anything the estimates imply that Black applicants are penalized as the share of Black children increases. Interestingly, the last cluster of coefficients, testing for Hispanic customer discrimination, shows evidence in favor of such discrimination, unlike the results in Table 3. Specifically, program directors strongly favor Hispanic applicants over their White counterparts as the share of Hispanic children increases (by a statistically significant 31.8 to 37.6 percentage points). In addition, a comparison of Black × Hispanic_share and Hispanic × Hispanic_share implies that Hispanic applicants are also favored over their Black counterparts.

In sum, our results indicate that customer discrimination may play a role in teacher hiring decisions. We show that neighborhood racial and ethnic shares—the most commonly used proxy for firms’ customer base—primarily advantage White and Black job seekers. As the fraction of children of a given race or ethnicity increases within a neighborhood, job seekers with the same background are advantaged in the hiring process. Results for the program-level racial and ethnic shares suggest this is also the case for programs whose enrollees are increasingly White and Hispanic. Black applicants, on the other hand, do not benefit as the number of Black children within a program increases. Given the strong correlations between the program- and census tract–level population shares, the lack of full consistency in our results requires some discussion.¹⁴ It is possible that for some demographic groups, the local racial-ethnic share is a noisy proxy for a firm’s actual customer base. For Hispanics, this seems plausible given that such children are less likely to participate in center-based care than their White and Black counterparts.¹⁵ Instead, Hispanics are more likely to rely on parental care or informal modes of nonparental care. Thus it follows that a neighborhood’s Hispanic share will not be a good proxy for the customer base if these families are less likely to patronize local child care centers.

The Impact of Child Care Regulations

The final empirical section examines to what extent states’ child care regulations influence the racial and ethnic gap in interview requests. These state-level policies are aimed at mitigating the child care market’s information problems by ensuring that a minimum level of quality is achieved across the spectrum of center-based programs. A key goal of child care regulations is to compel providers to improve the quality of their teacher workforce. As a result, it is possible that such policies increase the cost of discriminating against well-qualified minority applicants. One hypothesis, therefore, is that center-based providers operating in states with more stringent regulations are more likely to signal interest in minority applicants than their counterparts operating in less stringent policy environments.

Regulations cover virtually all dimensions of center-based programs. Our focus in this article is to study the impact of regulations governing the labor intensiveness of child care provision and staff qualifications. In particular, we examine child-to-staff ratios, maximum group sizes, and experience and education requirements for assistant teachers, lead teachers, and program directors. We selected these domains because they are among the most commonly studied in regulation literature, they are known to be correlated with measures of overall child care quality, and they are likely to be given significant consideration during the teacher-hiring process. We obtained information on states’ regulations from the National Database of Child Care Licensing Regulations (maintained by the U.S. Department of Health and Human Services) and Lieberman (2017). We code these regulations for the 17 states represented in the résumé audit study.¹⁶

There are two complications involved in coding these regulations. First, as others point out, individual regulations are highly correlated with one another (Blau, 2003; Hotz & Xiao, 2011). Second, states tend to enact different regulations aimed at children of different ages. For example, states vary the stringency of child-to-staff ratios by age, with the lowest ratios set for infants and toddlers and higher ratios set for preschool-age children. Such a large and highly correlated set of variables makes it difficult to empirically identify the impact of individual regulations. Therefore, we handle these complications by first standardizing the values of each regulation to have a mean of 0 and a standard deviation of 1. We then calculate the average of the domain-specific regulations so that we have an index of regulatory stringency for each major domain (e.g., child-to-staff ratios). We also construct an overall index of regulatory stringency by averaging over the primary domains. Increasing values for a given standardized index represents increasing regulatory stringency. Summary statistics for the individual regulations in each domain are presented in online Appendix Table 6.

The Impact of Regulations on the Racial and Ethnic Gap in Interviews

To estimate the impact of states’ child care regulations on the racial and ethnic gap in interview requests, we estimate models of the following form:

\begin{array}{l} {Interview}_{i k c m} = β_{0} + β_{1} regulation + β_{2} (Black \times regulation) \\ + β_{3} (Hispanic \times regulation) \\ + X' β + α_{c} + δ_{m} + λ_{k} + ε_{i k c m,} \end{array}

(4)

where regulation is one of the domain-specific indices of regulatory stringency (e.g., child-to-staff ratios) as well as the overall index of regulatory stringency. The other variable and coefficient vectors are identical to those described in Equation (1a). To estimate this model, we interact a given child care regulation variable with the indicators for Black and Hispanic and include these interactions along with regulation in the regression. Suppressed from the model are the main effects on Black and Hispanic so that the coefficient on the interactions can be interpreted as the percentage-point change in the likelihood that a Black or Hispanic applicant receives an interview given a one-standard-deviation increase in the stringency of regulation. Once again, we utilize this approach in order to understand whether, and by how much, these state-level regulations influence the Black–White and Hispanic–White interview gaps. We estimate a version of this model that replaces the city and month fixed effects with job advertisement fixed effecs.¹⁷

Results from Equation (4) are presented in Table 4. Each column shows the estimates on β₁, β₂, and β₃ for each regulatory domain. Panel A provides the OLS estimates, and Panel B provides the FE model estimates; both sets of results are quite similar. Overall, the results strongly suggest that increasing the stringency of states’ child care regulations reduces the racial and ethnic gap in interview requests. For example, estimates in column (1) of Panel A imply that Black applicants are 3.4 percentage points more likely to receive an interview request for each one-standard-deviation increase in the strictness of states’ child-to-staff ratios. Comparable estimates for Hispanics are between 1.1 (Panel A; not statistically significant) and 1.8 (Panel B; statistically significant) percentage points. These positive effects apply to regulations for maximum group size and lead teachers’ experience and education, whereas there are null effects of the assistant teacher and program director requirements. As shown in column (6), which presents the coefficients for the full regulation index, Black and Hispanic applicants are increasingly favored as the broad regulatory landscape becomes stricter. A one-standard-deviation increase in the index increases the likelihood of receiving an interview request by 4.3 percentage points among Blacks and increases Hispanics’ likelihood of receiving an interview request between 1.6 (Panel A; not statistically significant) and 2.8 (Panel B; statistically significant) percentage points.

Table 4

The Impact of States’ Child Care Regulations on the Racial and Ethnic Gap in Interviews

	(1)	(2)	(3)	(4)	(5)	(6)
Variable	Staff-to-Child Ratio	Maximum Group Size	Assistant Teacher Reqs.	Lead Teacher Reqs.	Program Director Reqs.	Full Regulation Index
Panel A: OLS estimates
Regulation	−.0380* (.0224)	−.0229 (.0141)	.0175 (.0471)	−.0203 (.0165)	−.0026 (.0292)	−.0367 (.0247)
Black × Regulation	.0338*** (.0108)	.0368*** (.0109)	−.0078 (.0246)	.0227* (.0120)	−.0023 (.0158)	.0434*** (.0145)
Hispanic × Regulation	.0111 (.0105)	.0151 (.0115)	−.0206 (.0246)	.0082 (.0125)	.0058 (.0159)	.0161 (.0146)
Panel B: Job advertisement FE estimates
Regulation	—	—	—	—	—	—
Black × Regulation	.0304*** (.0103)	.0355*** (.0099)	−.0294 (.0232)	.0240** (.0111)	.0001 (.0143)	.0415*** (.0134)
Hispanic × Regulation	.0175* (.0096)	.0199* (.0104)	−.0246 (.0230)	.0189* (.0111)	.0235 (.0143)	.0279** (.0132)
Observations	10,986	10,986	2,712	8,591	10,986	10,986

Note. Standard errors (in parentheses) are clustered at the job advertisement level. Panel A estimates the models using ordinary least squares and includes the résumé characteristics, city and month fixed effects, and résumé order indicators. Panel B includes job advertisement fixed effects. The fixed-effects models omit the city and month fixed effects and include the job advertisement fixed effects. The regulation estimated in each column is, respectively, the index of staff-to-child ratios, the index of maximum group size, the index of experience and education requirements for assistant teachers, the index of experience and education requirements for lead teachers, the index of experience and education requirements for program directors, and the overall index of regulatory stringency. The models in columns (1), (2), (5) and (6) are estimated on the full set of résumés, whereas those in columns (3) and (4) are estimated on the subset of résumés submitted to assistant and lead teacher positions, respectively. Reqs. = requirements.

p < .10. **p < .05. ***p < .01.

The preceding results indicate that child care regulations reduce the racial and ethnic disparity in interview requests, perhaps because heavily regulated providers face a higher cost of discriminating against well-qualified minorities. If this is the case, then one would expect highly qualified Black and Hispanic applicants to benefit more from these regulations than their less qualified counterparts. In other words, if regulations increase the incentive for providers to focus less on applicants’ race and ethnicity and more on their credentials, then highly qualified minorities should benefit disproportionately from this shift. Analyses in online Appendix Table 7 test this proposition by estimating Equation (4) on subsets of low- and high-quality résumés submitted for lead teacher positions.¹⁸ Looking at Panel A (i.e., low-quality résumés), we find that child care regulations do not advantage less qualified minorities: The coefficients on the interaction terms are small in magnitude and sometimes negatively signed, and none are statistically significant. In contrast, as shown in Panel B (i.e., high-quality applicants), regulations have sizable positive effects on interview requests for high-quality minorities. Several of the individual interactions are statistically significant, and tests of the equality of the interactions across low- and high-quality applicants often reject the null hypothesis (Panel C).

Finally, online Appendix Table 8 provides results for subsets of child care providers located in different markets. Specifically, we divide the sample by zip code median household income, thereby allowing us to test for differential effects of regulations across providers located in low- and high-income markets.¹⁹ This analysis is motivated by the findings in Hotz and Xiao (2011), who show that although regulations improve child care quality, these benefits accrue disproportionately to providers located in higher-income markets. One might expect, then, that the reduction in the racial and ethnic interview gap is also pronounced for providers in wealthy markets relative to those located in lower-income areas. Our results strongly suggest this is the case. Looking at Panel A (i.e., the low-income zip codes), we find that providers located in low-income markets do not respond to regulations by interviewing (relatively) more Blacks and Hispanics. If anything, the estimates imply that providers are less inclined to interview minorities. On the other hand, the estimates in Panel B (i.e., the high-income zip codes) reveal that providers located in wealthy markets increasingly favor minority applicants as the regulatory environment becomes stricter. Several of the individual interactions are statistically significant, particularly for Blacks, and tests of the equality of the interactions across low- and high-income zip codes often reject the null hypothesis (Panel C).

Summary and Discussion of Results

The goal of this article is to shed light on three dimensions of racial and ethnic discrimination in the labor market for center-based child care teachers: (a) to what extent discrimination exists in the hiring process, (b) whether customer discrimination plays a role in hiring behavior, and (c) to understand the impact of states’ child care regulations on the treatment of minority job seekers. In this section, we summarize the key results from our analyses and discuss how they inform the child care and discrimination literatures.

Our first set of analyses showed that Black and Hispanic child care job seekers are less likely to receive an interview than otherwise identical White applicants. Furthermore, these interview gaps persist across lower-skilled (e.g., assistant teacher) and higher-skilled (e.g., lead teacher) positions. We also uncover evidence that program directors’ own-race preferences may influence teacher-hiring behavior. In particular, we find that White directors favor White applicants and minority directors favor those from their own racial and ethnic background.

Several insights stem from these results. First, that the interview gap applies to applicants for lower- and higher-skilled positions may be interpreted as suggestive evidence of customer discrimination. Indeed, a key responsibility for child care teachers at all levels is to interact with parents about highly sensitive issues. It is therefore possible that center directors are responsive to parents’ racial preferences when hiring both beginner and advanced teachers. Second, the results for directors’ own-race preferences add important nuance to the way in which racial discrimination operates in the child care market: Although in the aggregate, White applicants are favored over their Black and Hispanic counterparts, the race and ethnicity of the program director clearly matters. Thus our results provide one explanation for why the child care teacher workforce remains less diverse than the children it serves: Program directors are overwhelmingly White. Such findings imply that one way to increase the diversity of the teacher workforce is to increase the diversity of program directors.

The second set of analyses was devoted to understanding whether customer discrimination is present in the labor market for child care teachers. Such an analysis has been difficult to undertake in traditional résumé audit studies because of difficulties collecting data on the racial composition of firms’ customer base. First, we analyze the impact of the potential customer base, as measured by the racial composition of the surrounding neighborhood, which has been the predominant approach in the literature. Second, our study contributes to this strand of the discrimination literature by analyzing the racial composition of the actual customer base, as measured by the children enrolled in the centers. Both sets of analyses uncover evidence consistent with customer discrimination: As the fraction of children of a given race or ethnicity increases within a neighborhood or program, job seekers with the same background are advantaged in the hiring process. This appears to be particularly true for child care programs whose potential and actual customers are increasingly White.

Given the well-known information asymmetries in the child care market, it is perhaps not surprising that our analysis uncovered evidence of customer discrimination. In the absence of being fully informed about the determinants of child care quality, parents may instead rely on the easy-to-observe features of the program setting, which includes the staff racial and ethnic composition. This leaves open the possibility that parents use teachers’ observable demographic characteristics to make inferences about their unobserved productive capacity. It is also possible that parents rely on these observable characteristics to evaluate other teacher traits, such as trustworthiness and empathy. Our results imply that parents rate teachers of their own race and ethnicity more highly on one or more of these dimensions. Insofar as parents use teachers’ race and ethnicity to make these judgments, child care directors appear to respond by favoring job seekers whose own background best matches that of the children attending the center.

The final set of analyses examined the impact of states’ center-based child care regulations on race-based interview decisions. We show that providers exposed to stringent regulations are more likely to interview minority applicants than their counterparts in comparatively lenient policy environments. We also find that these labor market advantages accrue disproportionately to high-quality minority applicants and to those applying to child care providers located in high-income communities. A key feature of our results is that with the exception of assistant teacher positions, the benefits apply to a variety of teacher types (e.g., infant, toddler, and preschool teachers). A potential explanation for the null effects on assistant teachers is that fewer states regulate experience and education levels for these entry-level positions, and among states that do, the regulations are less demanding than those aimed at lead teachers. Therefore, it is possible that child care providers find it feasible to engage in discrimination in the assistant teacher market—where the regulations are lighter—but infeasible to do so in the market for lead teachers—where the regulations are stricter.

It is worth noting that our regulation results are relevant to two distinct though related literatures. First, our research contributes directly to the literature studying the impact of regulations on the child care market (e.g., Blau, 2001, 2003; Hotz & Kilburn, 1994; Hotz & Xiao, 2011). The evidence suggests that increasing the stringency of regulations has small negative effects on supply and positive effects on prices and quality. Results in Hotz and Xiao’s (2011) analysis are especially noteworthy: Regulations reduce supply in low-income neighborhoods, where consumers respond by shifting into lower-quality informal arrangements, whereas they increase quality particularly in high-income communities. Our results shed light on a possible mechanism through which regulations improve provider quality in high-income communities: by reducing disparities in the treatment of high-skilled minority applicants during the hiring process. Second, our work informs the literature on the influence of asymmetric information on Black–White employment disparities. Some of this work focuses on the effect of occupational licensing and “ban-the-box” policies on labor market outcomes (e.g., Agan & Starr, 2017; Blair & Chung, 2017; Doleac & Hansen, 2016). Whereas this research focuses on whether occupational licensing eliminates the information gap between employers and applicants, our work studies the role of regulations in reducing information problems between consumers and producers. In the former case, minority job seekers are affected directly from the reduction in statistical discrimination, but in our case, one may argue that minorities are affected indirectly from firms’ increased emphasis on applicants’ credentials as opposed to their race and ethnicity.

Supplemental Material

EdR867941_Boyd-SwanandHerbst_OnlineAppendix – Supplemental material for Racial and Ethnic Discrimination in the Labor Market for Child Care Teachers

Supplemental material, EdR867941_Boyd-SwanandHerbst_OnlineAppendix for Racial and Ethnic Discrimination in the Labor Market for Child Care Teachers by Casey Boyd-Swan and Chris M. Herbst in Educational Researcher

Footnotes

ORCID iD

Casey Boyd-Swan

Notes

Authors

CASEY BOYD-SWAN, PhD, is an assistant professor in the Department of Political Science at Kent State University, 302 Bowman Hall, Kent, OH, 44242; cboydswa@kent.edu . Her research focuses on public policy issues related to parental employment, child care labor markets, and child and family well-being.

CHRIS M. HERBST, PhD, is an associate professor in the School of Public Affairs at Arizona State University, 411 N. Central Avenue, Phoenix, AZ, 85004; chris.herbst@asu.edu . His research focuses on the evaluation of child care markets and policies as well as the early childhood education workforce.

References

Agan

Starr

(2017). Ban the box, criminal records, and racial discrimination: A field experiment. Quarterly Journal of Economics, 133(1), 191–235.

Bar

Zussman

(2017). Customer discrimination: Evidence from Israel. Journal of Labor Economics, 35, 1031–1059.

Bartoš

Bauer

Chytilová

Matějka

(2016). Attention discrimination: Theory and Field experiments with monitoring information acquisition. American Economic Review, 106, 1437–1475.

Bassok

Markowitz

A.J.

Player

Zagardo

(2018). Are parents’ ratings and satisfaction with preschools related to program features? AERA Open, 4(1).

Becker

(1957). The economics of discrimination. Chicago, IL: University of Chicago Press.

Bertrand

Mullainathan

(2004). Are Emily and Greg more employable than Lakisha and Jamal? A field experiment on labor market discrimination. American Economic Review, 94, 991–1013.

Blair

Chung

(2017). Occupational licensing reduces racial and gender wage gaps: Evidence from the Survey of Income and Program Participation (HCEO Working Paper No. 2017-050). Chicago, IL: Human Capital and Economic Opportunity Global Working Group.

Blau

(2001). The child care problem: An economic analysis. New York, NY: Russell Sage Foundation.

Blau

(2003). Do child care regulations affect child care and labor markets? Journal of Policy Analysis and Management, 22, 443–465.

10.

Boyd-Swan

Herbst

C. M.

(2017a). The demand for teacher characteristics in the market for child care: Evidence from a field experiment (IZA Discussion Paper No. 10702). Bonn, Germany: Institute of Labor Economics.

11.

Boyd-Swan

Herbst

C. M.

(2017b). Technical appendix: The demand for teacher characteristics in the market for child care: Evidence from a field experiment. http://www.chrisherbst.net/files/Download/C._Herbst_RAS_TA.pdf

12.

Boyd-Swan

Herbst

C. M.

(2018). The demand for teacher characteristics in the market for child care: Evidence from a field experiment. Journal of Public Economics, 159, 183–202.

13.

Caldera

Y. M.

Hart

(2004). Exposure to child care, parenting style, and attachment security. Infant and Child Development, 13, 21–33.

14.

Carlsson

Rooth

(2007). Evidence of ethnic discrimination in the Swedish labor market using experimental data. Labour Economics, 14, 716–729.

15.

Chaudry

Pedroza

Sandstrom

Danziger

Grosz

Scott

M.M.

Ting

(2011). Child care choices of low income working families. Washington, DC: Urban Institute.

16.

Clark

Anderson

Adams

Baker

(2008). Assessing the knowledge, attitudes, behaviors and training needs related to infant feeding, specifically breastfeeding, of child care providers. Maternal and Child Health Journal, 12, 128–135.

17.

Corcoran

Steinley

Grady

(2017). Early childhood program participation, results from the National Household Education Surveys Program of 2016 (NCES 2017-101). Washington, DC: National Center for Education Statistics, Institute of Education Sciences, U.S. Department of Education.

18.

Dee

T. S.

(2004). Teachers, race, and student achievement in a randomized experiment. Review of Economics and Statistics, 86, 195–210.

19.

Dee

T. S.

(2005). A teacher like me: Does race, ethnicity, or gender matter? American Economic Review, 95, 158–165.

20.

Doleac

J. L.

Hansen

(2016). Does “ban the box” help or hurt low-skilled workers? Statistical discrimination and employment outcomes when criminal histories are hidden (NBER Working Paper No. 22469). Cambridge, MA: National Bureau of Economic Research.

21.

Downer

Goble

Myers

Pianta

(2016). Teacher-child racial/ethnic match within pre-kindergarten classrooms and children’s early school adjustment. Early Childhood Research Quarterly, 37, 26–38.

22.

Egalite

A. J.

Kisida

(2018). The effects of teacher match on students’ academic perceptions and attitudes. Educational Evaluation and Policy Analysis, 40(1), 59–81.

23.

Egalite

A. J.

Kisida

Winters

M. A.

(2015). Representation in the classroom: The effect of own-race teachers on student achievement. Economics of Education Review, 45, 44–52.

24.

Fryer

Levitt

(2013). Testing for racial differences in the mental ability of young children. American Economic Review, 103, 981–1005.

25.

Gershenson

Hart

Constance

Papageorge

(2017). The long-run impacts of same-race teachers (IZA Discussion Paper No. 10630). Bonn, Germany: Institute of Labor Economics.

26.

Gershenson

Holt

S. B.

Papageorge

N. W.

(2016). Who believes in me? The effect of student–teacher demographic match on teacher expectations. Economics of Education Review, 52, 209–224.

27.

Gilliam

W. S.

Maupin

A. N.

Reyes

C. R.

Accavitti

Shic

(2016). Do early educators’ implicit biases regarding sex and race relate to behavior expectations and recommendations of preschool expulsions and suspensions? Research study brief, Yale University Child Study Center, New Haven, CT.

28.

Glazerman

Dotter

(2017). Market signals: Evidence on the determinants and consequences of school choice from a citywide lottery. Educational Evaluation and Policy Analysis, 39, 593–619.

29.

Graves

Howes

(2011). Ethnic differences in social-emotional development in preschool: The impact of teacher child relationships and classroom quality. School Psychology Quarterly, 26, 202–214.

30.

Holt

S. B.

Gershenson

(2015). The impact of teacher demographic representation on student attendance and suspensions (IZA Discussion Paper No. 9554). Bonn, Germany: Institute of Labor Economics.

31.

Holzer

Ihlanfeldt

(1998). Customer discrimination and employment outcomes for minority workers. Quarterly Journal of Economics, 113, 835–867.

32.

Hotz

V. J.

Kilburn

(1994). Regulating child care: The effects of state regulations on child care demand and its costs. Working paper, Rand Corporation, Santa Monica, CA.

33.

Hotz

V. J.

Xiao

(2011). The impact of regulations on the supply and quality of care in child care markets. American Economic Review, 101, 1775–1805.

34.

Joshi

Ocanto

Jacobs

R. J.

Bhoopathi

(2016). Florida child care center directors’ intention to implement oral health promotion practices in licensed child care centers. BMC Oral Health, 16, 100.

35.

Kids Count Data Center. (2017). Child population by race and age group. Baltimore, MD: Annie E. Casey Foundation.

36.

Lahey

Beasley

(2009). Computerizing audit studies. Journal of Economic Behavior and Organization, 70, 508–514.

37.

Leonard

Levine

Giuliano

(2010). Customer discrimination. Review of Economics and Statistics, 92, 670–678.

38.

Lieberman

(2017). A tale of two pre-K leaders: How state policies for center directors and principals leading pre-k programs differ, and why they shouldn’t. Washington, DC: New America Foundation.

39.

Lindsay

C. A.

Hart

C. M.

(2017). Exposure to same-race teachers and student disciplinary outcomes for black students in North Carolina. Educational Evaluation and Policy Analysis, 39(3), 485–510.

40.

Lutzkanin

K. M.

Myers

A. K.

Schaefer

E. W.

Sekhar

D. L.

(2016). Report of sugar-sweetened beverages offered in Pennsylvania childcare centers. Clinical Pediatrics, 55, 518–524.

41.

Magnuson

Waldfogel

(2005). Early childhood care and education: Effects on ethnic and racial gaps in school readiness. Future of Children, 15, 169–196.

42.

Mamedova

Redford

(2015). Early childhood program participation, from the National Household Education Surveys Program of 2012 (NCES 2013-029.REV). Washington, DC: National Center for Education Statistics, Institute of Education Sciences, U.S. Department of Education.

43.

Mocan

(2007). Can consumers detect lemons? An empirical investigation of information asymmetry in the market for child care. Journal of Population Economics, 20, 507–531.

44.

Murray

K. M.

Waas

G. A.

(2008). Child and teacher reports of teacher–student relationships: Concordance of perspectives and associations with school adjustment in urban kindergarten classrooms. Journal of Applied Developmental Psychology, 29, 49–61.

45.

National Survey of Early Care and Education Project Team. (2014). Household search for and perceptions of early care and education: Initial findings from the National Survey of Early Care and Education (NSECE) (OPRE Report No. 2014-55a). Washington, DC: Office of Planning, Research and Evaluation, Administration for Children and Families, U.S. Department of Health and Human Services.

46.

Neumark

Bank

Van Nort

(1996). Sex discrimination in restaurant hiring: An audit study. Quarterly Journal of Economics, 111, 915–941.

47.

Nunley

Pugh

Romero

Seals

(2014). An examination of racial discrimination in the labor market for recent college graduates: Estimates from the field. Auburn University Department of Economics Working Paper Series (AUWP 2014-06). Retrieved from https://cla.auburn.edu/econwp/archives/2014/2014-06.pdf

48.

Oreopoulos

(2011). Why do skilled immigrants struggle in the labor market? A field experiment with thirteen thousand resumes. American Economic Journal: Economic Policy, 3, 148–171.

49.

Park

McHugh

Batalova

Zong

(2015). Immigrant and refugee workers in the early childhood field: Taking a closer look. Washington, DC: Migration Policy Institute.

50.

Pungello

E. P.

Kurtz-Costes

(1999). Why and how working women choose child care: A review with a focus on infancy. Developmental Review, 19, 31–96.

51.

Risica

P. M.

Amin

Ankoma

Lawson

(2016). The food and activity environments of childcare centers in Rhode Island: a directors’ survey. BMC Nutrition, 2(1), 41.

52.

Saft

Pianta

(2001). Teachers’ perceptions of their relationships with students: Effects of child age, gender, and ethnicity of teachers and children. School Psychology Quarterly, 16, 125–141.

53.

Schaack

D. D.

V. N.

(2017). Colorado Early Childhood Workforce Survey 2017. Denver: University of Colorado Denver.

54.

Scheunemann

Schwab

Margaritis

(2015). Oral health practices of state and non-state-funded licensed childcare centers in Wisconsin, USA. Journal of International Society of Preventive & Community Dentistry, 5(4), 296.

55.

Schneider

Buckley

(2002). What do parents want from schools? Evidence from the Internet. Educational Evaluation and Policy Analysis, 24, 133–144.

56.

Shapiro

(1986). Investment, moral hazard, and occupational licensing. Review of Economic Studies, 53, 843–862.

57.

Sigman-Grant

Christiansen

Branen

Fletcher

Johnson

S.L.

(2008). About feeding children: Mealtimes in child-care centers in four western states. Journal of the American Dietetic Association, 108(2), 340–346.

58.

Sigman-Grant

Christiansen

Fernandez

Fletcher

Branen

Price

B. A.

Johnson

S. L.

(2011). Peer reviewed: Child care provider training and a supportive feeding environment in child care settings in 4 states, 2003. Preventing Chronic Disease, 8(5), A113.

59.

Tenenbaum

H. R.

Ruck

M. D.

(2007). Are teachers’ expectations different for racial minority than European American majority students? A meta-analysis. Journal of Educational Psychology, 99, 253–273.

60.

U.S. Census Bureau. (2012). Most children younger than age 1 are minorities, Census Bureau reports [Press release]. Washington, DC: Author.

61.

Wright

Gottfried

(2017). A kindergarten teacher like me: the role of student-teacher race in social-emotional development. American Educational Research Journal, 54, 78S–101S.

62.

Zussman

(2013). Ethnic discrimination: Lessons from the Israeli online market for used cars. Economic Journal, 123, F433–F468.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.22 MB