Public Childcare

The paper estimates net costs for two groups of beneficiaries: compliers and always-takers. Compliers are those who participate in the childcare program only if assigned to the treatment group (that is, individuals who would not have participated in childcare if not for the program). Always-takers are those who would participate in a childcare program regardless of the program.

The paper estimates two direct cost components for compliers. First, they calculate routine program expenses at approximately US$92 per child annually. To estimate the total program cost, they divide this amount by the number of centers to determine the average cost per center. They then factor in the empirical occupancy rate per center to arrive at a per-child cost, which they estimate to be $92. Second, infrastructure costs are imputed as US$55 per child per year. When combined, the total cost of the program for compliers was US$92 + US$55 = US$147 per child per year (US$159 for 13 months).

The paper assumes a program cost of zero for always-takers because when they move to a new childcare center that opened as a result of Programa Urbano, they open a slot in their old center. In other words, the paper assumes the effect of the program on always-takers is to shift them from a center that is farther away to a closer center.

As compliers are 46% of the sample population and the estimated program cost for always-takers is zero, the average direct cost of the program is US$159 x 46% = US$73.

The paper also estimates two revenue externalities associated with the compliers. First, the paper calculates US$31 of higher income tax revenues from increased maternal earnings. The paper found that mothers were 12% more likely to work because of the program. Assuming minimum wage level of US$1,618 a year and adjusting for the length of the program, compliers generate 12% x US$1,618 x 15% tax rate x 13 months/12 months = US$31 of revenue externalities.

Second, the paper estimates that the increase in tax revenues from higher lifetime earnings for children due to improved socio-emotional skills is US$117. The paper finds that the program improves socio-emotional skills among children by 0.38 standard deviations. Utilizing results from Lindqvist and Vestman (2011), the paper estimates that a one standard deviation increase in noncognitive skills could have an impact of 7.4 percent on log wages, suggesting the program increase of 0.38 standard deviations could lead to an increase in increase lifetime earnings of 2.8 percent. Assuming the children earn minimum wage, the program’s impact on children’s lifetime earnings amounts to US$783 in present value, yielding tax revenue of approximately 15 percent tax rate x US$783 = US$117.

Together, the total fiscal externality is (US$31 + US$117) x 46% = US$69 once adjusting for the share of compliers in the population.

Thus, the net cost is program per child is US$159 – (US$31 + US$117) x 46% = US$5.

(\frac{ATR^1}{1-ATR^0})

The MVPF is then the ratio of the willingness to pay to the net cost: US$331/US$5 = 66 (which is reported in the paper as “around 70”).

The paper reports a few other MVPF values under different assumptions:

1. If they consider the actual interest rate paid by the program (1.2 percent), the MVPF becomes infinity.
2. If they consider a more conservative interest rate of 5 percent, the MVPF goes down to 6.8.
3. If they assume mothers and children earn 50% more than minimum wage, the MVPF becomes infinity.
4. If they increase the interest rate to 5 percent and increase wages to 50% more than minimum wage, the MVPF is 46.