Impacts of early childhood care subsidy designs on the care services market
Won Lee, University of Minnesota
Aaron Sojourner, University of Minnesota
Liz Davis, University of Minnesota

11/7/2019
2019 APPAM, Denver, CO

Motivation and relevance

  • ECE subsidies can increase human capital accumulation, especially for children from disadvantaged families.(Barnett, 2008; Chaparro & Sojourner, 2015; Duncan & Magnuson, 2013; Duncan & Sojourner, 2013; Heckman, 2006)
  • Many policymakers want to invest in ECE:
    • Uncertainty remains about how subsidy expansions affect crowd-out, prices, and other outcomes.
    • Less study of how ECE subsidy policies affect the ECE services market
  • Large variation in ECE subsidies in Minnesota in recent years

Research Question

  • To what extent does each additional subsidy of $10,000 per child in a community's population affect:
    • the quantity and types of care services
    • market prices of care services
    • differently if the subsidy target infants and toddlers (age 0-2) or preschoolers (age 3-5)
    • differently if the subsidy target public providers or private ones

Literature

  • Funding for child care subsidies have been associated with growth in the number of providers (Hatfield et al 2015; Cochi Ficano 2006; Kreader et al., 1999)
  • Only few but growing lierature on the system-level impacts:
    • Effects of universal prekindergarten
      • Georgia (voucher) and Oklahoma (school based) on capacity, ECE workforce and pay (Bassok, Fitzpatrick, Loeb, 2014)
      • Floridas rollout on county-level capacity (Bassok, Miller Galdo, 2016)
      • NYC rollout on hex-level capacity (Brown, 2018)

Data

  • Provider Data
    • Provider-year level data
      • 2012-2015 NACCRAware (DHS)
      • Capcity
      • Price
      • Location
      • Other characteristics (Care Type, Age-group served)
  • Policy Data
    • Locale-year level data
    • Various sources (no unified data system)
    • Multiple programs at different geographic-level
  • Analytic Data
    • Harmonize to the school-district level

Measures

  • Outcome
    • Total quantity
      • By provider type (Center, Family, and Public), by age group (infant, toddler, and preschool)
    • Average Price
      • By provider type (Center, Family, and Public), by age group (infant, toddler, and preschool)
    • Share highly-rated under the state's quality rating and information system (3 or 4 stars in Parent Aware QRIS system)
  • Treatment
    • Policy dollar amounts
      • $ per child, if outcomes that are averages (prices)
      • $ level, if outcomes that are totals (capacity)

Policy Data Sources

Program Age 0-2 Age 3-5 Private Public Data Observed
Race to the Top (RTT) scholarships Y Y Y transformation zone
Early Learning Scholarship family (ELSF) Y Y super-county zone
Early Learning Scholarship provider (ELSP) Y Y school-district zone
Early Head Start Y Y provider location
Head Start Y Y provider location
Child Care Development Fund Y Y Y residence location
Voluntary Prekindergarten Y Y residence location
All-Day Kindergarten Y Y provider location
Note: Public and rectricted data comes from either Minnesota Department of Education or Department of Human Services

Descriptive characteristics of policy variables (School district-year level, 2012-2015)

Mean Std. Dev.
All funding 74.715 236.621
Private-provider funding 44.139 148.663
Private-young funding 14.178 46.136
Private-old funding 29.961 103.184
Public-provider funding 30.576 97.767
Public-young funding 5.617 19.731
Public-older funding 24.959 80.681
AllDayK funding 101.531 220.587
Funding/kid 0.125 0.215
Private-prov funding/kid 0.065 0.084
Private-young funding/kid 0.048 0.132
Private-older funding/kid 0.049 0.082
Public-prov funding/kid 0.061 0.167
Public-young funding/kid 0.048 0.132
Public-older funding/kid 0.013 0.063
AllDayK funding/kid 0.097 0.076
Young funding/kid 0.063 0.132
Older funding/kid 0.062 0.117
Note: The policy per children under age 5 variables in dollar ($10,000).; The policy variables in dollar ($10,000) ; N=1191

Descriptive characteristics of age-specific outcome variables (School-District level)

Mean Std. Dev.
Infant
Capacity 180.515 373.73
Family cap. 64.287 105.223
Center cap. 42.283 136.12
Public cap. 73.945 171.31
High-rated cap. 20.147 66.17
Price 126.043 32.527
Center price 198.921 72.649
Family price 122.413 26.382
Toddler
Capacity 250.099 520.511
Family cap. 101.599 165.054
Center cap. 74.556 238.422
Public cap. 73.945 171.31
High-rated cap. 34.173 109.518
Price 120.166 28.67
Center price 174.514 58.731
Family price 117.462 24.149
Preschool
Capacity 554.304 1055.928
Family cap. 317.444 497.683
Center cap. 162.882 498.362
Public cap. 73.978 171.368
High-rated cap. 78.199 245.442
Price 116.664 24.943
Center price 157.652 49.142
Family price 114.41 21.342
Note: N=1191

Methods


Difference-in-differences model:

\begin{align*} y_{lt} = \beta P_{lt} + X_{lt}\gamma + \gamma_{l} + \rho_{t} + \varepsilon_{lt} \end{align*}


  • $y_{lt}$ represents an quanity and price outcomes
  • $P_{lt}$ represents the policy dollar amount of interest, ten thousands of dollars per local child
  • $X_{lt}$ is a control variable for locale-year all-day kindergarten funding.
  • $\gamma_{l}$ is SD fixed effects
  • $\rho_{t}$ is year fixed effects locales.
  • $\varepsilon_{lt}$ measure locale-year specific unobservable influence.

Results

Estimated effects of overall funding and sector-specific funding on capacity

Results

Estimated effects of overall funding and sector-specific funding on private prices

Future analysis

  • Improve accuracy of the data
    • Better policy data
      • Voluntary Prekindergarten
      • ELS: Finer granular data
    • More years of outcome data
      • 2016, 2017, 2018
  • More age-group specific analysis
Thank you!