Blau and Currie 2004 — Preschool Day Care and Afterschool Care

Summary

A 126-page comprehensive review chapter for the Handbook on the Economics of Education (NBER Working Paper 10670) covering the economics of preschool, day care, and afterschool care. The paper is organized around two interlinked literatures: the market for private child care (demand, supply, quality, and government subsidies) and the effectiveness of publicly provided early intervention programs (Head Start, model intensive programs, state pre-K, and afterschool care). The central policy tension running through the review is the conflict between using child care policy to increase maternal employment versus using it to improve child development outcomes — goals that call for different subsidy designs and often point in opposite directions.

Key Claims

Who Uses Child Care

Female labor force participation of married mothers with children under 6 rose from $11.9\%$ in 1950 to $62.8\%$ in 2000; non-maternal care is now the norm, with most infants in non-parental care by four months.
Even one-third of children of non-employed mothers are in non-parental child care, indicating child care is not purely a labor-supply complement.
$10.5\%$ of 5–14 year olds of employed mothers are in unsupervised self-care; the probability rises steeply with age ( $8.1\%$ at age 9, $18.1\%$ at age 11, $44.9\%$ at age 14). Juvenile crime triples in the 3–6pm after-school window.

Demand: Price Elasticities

Twenty studies estimate the employment elasticity with respect to child care price; estimates range from $-0.04$ to $-1.26$ in the mainstream binomial probit literature, and from $-0.09$ to $-0.20$ in the better-specified multinomial models that properly account for the unpaid care option.
The three most credible studies (Ribar 1995; Blau and Hagy 1998; Tekin 2002) all produce elasticities at the low end: $-0.09$ to $-0.20$ . "The best available estimates suggest that the effects of the price of paid child care on labor force participation, hours of work, and welfare use are small."
The key identification problem: child care price is measured from a selection-corrected expenditure equation, but since quality is a choice variable, there are no valid exclusion restrictions — identification relies on functional form.

The $\alpha$ vs. $\beta$ Subsidy Trade-off

Child care price can be decomposed as $p = \alpha + \beta q$ $p = α + β q$ (a fixed component $\alpha$ $α$ plus a quality premium $\beta q$ $β q$ ). This generates a sharp policy trade-off:
- An $\alpha$ -subsidy (unconditional on quality) maximally increases employment by reducing the full price of any care.
- A $\beta$ -subsidy (larger for higher-quality care) more effectively improves quality demanded.
These two goals are in direct conflict: the subsidy design that best promotes employment is the design worst for quality, and vice versa.

Supply and the Wage Puzzle

Child care labor supply elasticity $\approx$ $1.15$ (Blau 2001, Current Population Survey (CPS) 1977–1998).
Despite a $24\%$ increase in demand for child care 1983–1998, real child care wages rose only $8\%$ (vs. the predicted $17\%$ ). The shift toward center-based care (higher child-staff ratios) partly accounts for the wage dampening.
This implies child care workers are paid below their marginal product relative to what supply conditions would predict — sustained by low outside options and the intrinsic motivation of many workers.

Child Care Quality

Process quality (caregiver interactions) matters more for child development than structural quality (ratios, group sizes), but process quality has no nationally representative data — the most recent data are from 1990.
Average Early Childhood Environment Rating Scale (ECERS) quality score across the Cost, Quality, and Outcomes (CQO) Study and National Child Care Staffing Study (NCCSS): just under $4$ on a 1–7 scale ("mediocre"; halfway between "minimal" and "good"). Infant-toddler classrooms score well below preschool classrooms.
Non-profit centers score higher than for-profits; quality varies substantially by location (California/Connecticut/Boston $\approx 1$ point above North Carolina/Atlanta/Seattle).
Cost function (Blau and Mocan 2002): raising quality from "minimal" ( $3$ ) to "good" ( $5$ ) raises costs by only $11.2\%$ — quality improvement is cheap.
Price function: quality elasticity of price $= 0.13$ – $0.40$ across states — market modestly rewards quality.
The quality puzzle: if improving quality is inexpensive and the market rewards it, why is quality so uniformly mediocre? Evidence suggests providers place no weight on quality in their utility function beyond its profit effect; and parents may not be able to observe or are unwilling to pay for better quality.
The only robust predictor of process quality across studies is teacher training — staff-child ratios, group size, and wages predict structural quality but teacher training predicts process quality even within-center.

Government Subsidies

Child care subsidies have grown substantially since the 1970s (Dependent Care Tax Credit (DCTC), Child Care and Development Block Grant (CCDBG), Aid to Families with Dependent Children Child Care / Transitional Child Care (AFDC-CC/TCC)). The DCTC is non-refundable and thus provides no benefit to the lowest-income families.
Studies of CCDF (Child Care and Development Fund) subsidies on employment find significant but small effects, consistent with the small price elasticity of employment.
Crowding out of private-sector child care by Head Start and public subsidies is entirely unstudied — a major gap.
Subsidy pass-through to prices and take-up decisions of eligible families are also largely unstudied.

Model Early Intervention Programs

Seven model programs with randomized designs from Barnett (1995) and Karoly (1998) reviews:
- Perry Preschool (2 years, age 3–4, 1:6 ratio, all masters-level teachers, weekly home visits): at age 27, higher graduation rates, employment, and earnings; lower crime, welfare use. $\$7$ in social benefits per $\$1$ spent (Karoly et al. 1998).
- Carolina Abecedarian (birth to age 5, full-day, 1:3–1:6 ratio, 50 weeks/year): at age 21, $2\times$ more likely to have attended a 4-year college; present discounted value (PDV) of benefits $\$76{,}000$ vs. costs $\$34{,}600$ at $5\%$ discount rate. Treatment effects twice as large for children from the poorest/least-educated families.
- Early Training Project (age 4–5, part-day, home visits): special education use $5\%$ vs. $29\%$ for controls at age 12.
- Chicago Child-Parent Centers (preschool + enriched K–3 curriculum): reduced high school dropout $24\%$ ; $\$3.69$ in government savings per $\$1$ spent; effects grow with duration in the program.
Positive effects operate primarily through non-cognitive skills (impulse control, social adjustment, reduced grade retention, behavioral outcomes), not IQ — a key result given that early IQ gains often fade.
Non-randomized studies of similar programs frequently find insignificant or wrong-signed effects, confirming the importance of experimental design.

Head Start

Serves $\approx 800{,}000$ children ( $60\%$ of eligible 3–4 year olds); FY2001 appropriation $\$6.2$ B; cost $\approx \$5{,}021$ /child/year part-day, $\approx \$9{,}000$ for full-day/full-year.
No large-scale randomized trial of a typical Head Start program had ever been conducted as of 2004 (the national Head Start Impact Study was just beginning).
Currie and Thomas (1995) sibling comparison using the National Longitudinal Survey of Youth (NLSY):
- African-Americans: initial test score gains fade out by elementary school — attributed not to Head Start quality but to subsequent school quality: black Head Start children attend lower-quality schools than other black children. Stratifying by school quality largely eliminates the race gap in fade-out.
- White children: persistent gains in test scores and reductions in grade repetition.
Garces, Thomas, and Currie (2002) using Panel Study of Income Dynamics (PSID) adults: white Head Start participants more likely to graduate high school and attend college; black participants significantly less likely to be charged with a crime.
Currie (2001) estimate: short/medium-term benefits (grade retention, special education savings) recoup $40$ – $60\%$ of Head Start costs. If long-term effects are even $1/4$ as large as model programs, the program pays for itself.
Head Start costs $\approx 71\%$ as much per child-year as Perry Preschool, but is less intensive; the gap in documented effectiveness probably reflects this intensity difference.

Early Head Start and State Programs

Early Head Start (infants/toddlers, 1994): at age 3, higher cognitive development scores, less aggressive behavior, better sustained attention. Long-run effects unknown at time of writing.
State preschool programs ( $\approx 800{,}000$ children, comparable scale to Head Start): limited evidence; cognitive test score gains often fade, but effects on attendance, grade retention, and behavioral outcomes sometimes persist to middle school.
Block grant concern: if federal Head Start funds are converted to state block grants, states may reallocate to child care subsidies rather than educational programming, which may not improve child outcomes.

Afterschool Programs

After-school programs are underutilized ( $59\%$ average capacity); paradoxically, used primarily by young children whose parents want supervision, not by older at-risk adolescents.
Model tutoring programs (Howard Street, Memphis City Schools) show significant reading gains in randomized designs.
LA's Best (most-studied large program): mixed results; some positive GPA and school attitude effects, but sensitive to outlier exclusion and control group selection.
Average available afterschool program: no detectable effect on outcomes. "It is a leap to argue that the average available after school program has any effect on child outcomes, since the model programs appear to be significantly better than the typical program."

Concepts Introduced or Extended

Entities Mentioned

Quotes

"There is a clear tradeoff in subsidy policy between the goals of increasing employment and improving the quality of child care."

"The authors of the CQOS report refer to this level of quality as 'mediocre'... If raising the quality of child care is relatively inexpensive and well rewarded, then why is so much privately provided child care of low quality?"

"It is surprising that there has never been a large-scale, randomized trial of a typical Head Start program."

"The evidence in support of favorable long-term effects of public programs is less conclusive than the evidence showing positive effects of model programs, mostly because there have been very few well-designed studies of longer-term effects."

My Take

The paper's most durable contribution is probably the formal $\alpha$ -vs- $\beta$ subsidy trade-off and the quality-puzzle framing — both clarify why market and government provision produce systematically mediocre outcomes even when improvement is cheap. The Head Start fade-out analysis (Currie and Thomas 1995), which attributes black-white differences in fade-out to subsequent school quality rather than program quality, is particularly important for the SSI and early disadvantage literature: it suggests that the problem is not whether children get a good start, but whether that start is sustained by subsequent institutional quality. The paper was written just before the Head Start Impact Study's randomized results were available (published 2010), which substantially confirmed the fade-out pattern and qualified the long-run evidence. For the wiki, the most relevant connections are to the SSI Children's Program literature — specifically the hypothesis that disability-benefit receipt in childhood functions as a partial substitute for the kind of long-run investment model programs like Perry Preschool represent.