Effect of Child Care Regulations on Child Care Markets: Evidence from Policy Discontinuity at the Border
Won Lee, First 5 California
Aaron Sojourner, W. E. Upjohn Institute for Employment Research
Elizabeth E. Davis, University of Minnesota
Jonathan Borowsky, University of Minnesota

11/22/2024
2024 APPAM Research Conference, National Harbor, MD

Motivation and relevance 1

  • In 2019, almost 60% of children under age 6 attended one or more weekly nonparental care setting (USDOE, 2021)
    • States have adopted regulations that set minimum standards for child care facilities and programs
    • Structural features of care settings thought to be related to quality and safety, such as Staff-to-Child Ratio (SCR)
  • Policy relevance
    • Understanding the role of regulation on the market is important as it might impact supply and price of care
    • Increasing interest in expansion of child care by deregulation
  • SCR strictness impacts childcare establishments through opposing forces
    • #1: ↑ employment, ↑ service quality → ↑ demand for ECE ↑ Q
    • #2: ↑ service costs → ↓ supply curve, ↓ demand for ECE employment ↓ Q
    • #2a: ↑ service costs → ↓ affordability of ECE → ↑ alternative care ↓ Q
    • The combined effect of these two opposing forces requires empirical investigation

Motivation and relevance 2

  • Relatively few empirical studies
    • Hotz and Xiao (2011)
      • Establishment-level panel data on child care centers from 1987 to 1997, finds:
        • stricter state regulation of staff-child ratios led to a decrease in the number of child care center establishments particularly in poor communities.
        • improved quality of care proxied by the share of child care centers with national accreditation, particularly in higher income areas.

This study

  • Re-examine Hotz and Xiao (2011)’s results using more recent data and three different quasi-experimental designs
    • 1. Two-way fixed effects design based on the county-level data
      • 2011, 2014, 2017 County Business Patterns data on childcare industry establishments, employment, and payroll
    • 2. Border regression discontinuity based on small geographic unit data within 10 miles of the border line
      • U.S. childcare provider’s exact location from Dun & Bradstreet (D&B) and estimates SCR effects focusing on fineresolution differences across state borders
      • The effect of stricter regulations on the number of establishments, their payroll and number of employees
    • 3. Synthetic control based on state-level data
      • Aggregated from 2011 to 2021 data from County Business Patterns data on childcare industry establishments, employment, and payroll
      • Exact date of the policy change is known and uses a synthetic control group approach

Summary statistics for county-level analysis

County-year level analysis: Effect of staff-to-child ratio on N of employees per 100 families, N of establishments per 100 families, annual payroll per Employee, and staff turnover rate

County-year level analysis: Effect of staff-to-child ratio on N of establishment per 100 families, by income tercile

Illustration of zones near the state border with distance-based subzones

Constructed zones for analysis
Provider-level D&B data near state border
  • Zones were created using the "Build Balanced Zones" tool in ArcGIS (Genetic Algorithm); parameters: number of zones, size, etc.
  • Zones are adjacent to the state border, limited to a 10-mile distance from the border line. Each state-zone is further partitioned into 5 subzones (2-mile blocks).
  • Provider-level data from D&B is aggregated to each subzone for analysis.

Summary statistics for regression discontinuity design

RDD analysis: Discontinuity at the border (Staff to Child Ratio)

Large discontinuity in policy at the border

RDD analysis: Discontinuity at the border (Outcomes)

Not much difference in outcomes at the border

RDD analysis: Effect of staff-to-child ratio on N establishments per 100 families, N of employees per 100 families, and sales volume per 100 families, by income tercile

Synthetic control estimation - Louisiana policy change in 2015

Figure 1: Effect on N Establishments
Figure 2: Effect on N Employees

Conclusion

Estimated coefficient on Staff-Child Ratio (sign and significance) in Each Specification
This Study
Hotz & Xiao 2011 County Level State Border RDD Synthetic Control
N of establishments (-) S. (-) N.S. (-) N.S. (-) N.S.
N of Employment n/a (-) N.S. (-) N.S. (-) N.S.
N of establishments in low-income areas (-) S. (-) N.S. (-) N.S.
N of Employment in low-income areas (+) S.
Payroll per employee (-) N.S.
Staff-Turnover rate (+) N.S.

Note: S. indicates statistical significance at conventional levels; N.S., if otherwise.

  • Analysis of data from 2011–2017 shows limited evidence that SCR significantly impacts overall market outcomes
  • The signs of the estimates generally align with findings from Hotz and Xiao (2011), indicating that stricter policies increase service costs and reduce demand
  • RDD analysis reveals an increase in childcare employment in lower-income areas
  • Going forward, plan to add more years of data to increase power (requires identifying exact year of policy change)

Appendix: Variation in the policy variable (Staff to child ratio)

Appendix: Hypothetical spatial distributions of providers near state border
by effect of state regulation on supply location and population distribution