Published Duke study of North Carolina early childhood programs finds good evidence for spillover benefits for overall student achievement

The recently-published version of a Duke University study provides good evidence that high-quality early childhood programs have sizable spillover benefits for overall student achievement.  The research also suggests that these programs can have ratios of earnings benefits to costs that exceed 10 to 1. The study, by Duke professors Helen Ladd, Clara Muschkin, and Kenneth Dodge, looks at North Carolina’s Smart Start child care program and More at Four pre-K program. I previously did a blog post on an early version of this research.

Smart Start is a state of North Carolina program, begun in 1993, that provides state aid to county partnerships (or in some cases, multi-county partnerships) that supported a wide variety of initiatives to improve the quality of early childhood services. With encouragement from state regulations, most funds were devoted to child care services, both to fund child care subsidies and improve child care quality.

More at Four was a state of North Carolina program from 2001 to 2011 that supported full-day pre-K programs at age 4. It has since been renamed NC Pre-K.

The Duke research study uses a clever methodology that provides good evidence of causal effects of the program on student achievement. The research relies on the accidents of history and politics. Both the Smart Start program and the More at Four program were gradually rolled out geographically to different counties. In addition, funding levels have varied over time both statewide and by county. The research examines whether the level of Smart Start spending or More at Four spending in a child’s birth county during the relevant years (zero to 4 for Smart Start, age 4 for More at Four) help explain the child’s third grade test scores. The research also controls for other factors that might affect a child’s test scores, such as the education of the child’s mother, and demographic factors. Finally, the research controls for other county trends that might affect test scores.

The research finds sizable effects of spending on these programs in a child’s birth county on the child’s subsequent test scores.  Furthermore, the research finds that these effects are particularly strong for the at-risk children who are targeted by both programs. Test score effects are stronger for children whose mother does not have a high school degree.

The research implies extremely high ratios of earnings benefits to costs. The third grade test score effects can be used to predict future earnings effects. The research suggests that typical levels of Smart Start or More at Four spending as of 2009 would be sufficient to increase overall average future earnings, averaged over all four-year-olds in the county,   by 1.6% for Smart Start, and 2.9% for More at Four.

This might not sound like much. However, this increase is over an entire future career, from the first job until retirement.  Even if real wages do not increase in the future, average career earnings for Americans exceed $1.5 million, so even a 1% to 3% earnings boost is a huge amount of money.

Furthermore, this earnings boost is very large relative to costs. These simulations are for typical levels of funding as of 2009 for these two programs. This spending would average about $1,100 for More at Four for the typical 4-year old in the county – the funding per child actually in pre-K was much higher, but only 20% or so of children in most counties were in a state-funded pre-K program.  The funding per child for Smart Start was around $220 per year for 5 years, birth to age 4, or total spending that also is about $1,100 for the average 4-year old in the county.

To compare the future earnings boost to these cost figures, we need to adjust the future earnings figures for the time value of money – even if there were no inflation, we value a dollar today more than a dollar 20 years from now because we can save and invest a dollar today to earn interest, and thereby have more than a dollar in the future. If we adjust future dollars to their “present value” at age 4 using an appropriate interest rate, we conclude that Smart Start increased the “present value” of future earnings for all county 4-year olds by around $13,000.  The ratio of these earnings benefits to costs would be over 10 to 1 (11.7=$12,854 divided by $1,100).

Similarly, for More at Four, the program would be estimated to increase the “present value” of future earnings by about $24,000. This implies a ratio of earnings benefits to costs of over 20 to 1 (21.5=$23,690 divided by $1,100).

(Technical note: These calculations used estimates by Chetty et al. that imply that on average, each one percentile increase in 3rd grade test scores increases future earnings by 0.612%. I averaged the reading and math results from the Duke study to get average percentile boosts in test scores. The overall earnings figures are figures derived from the American Community Survey. I adjusted figures used in Bartik, Gormley, and Adelstein  for Tulsa to 2012 national prices.  I assumed that future real earnings grew by 1.2% per year, and adjusted future earnings to a present value at age 4 using a 3% real discount rate. The resulting present value of future earnings as of age 4 for an average child was $818,275.)

These estimates imply large spillover benefits of these programs for children other than those directly served by the programs. This is most clearly seen for the More at Four program, which only directly provides pre-K for about 20% of all children, with those services targeted on children from low-income families or with other risk factors.

As Ladd, Muschkin, and Dodge point out, their estimates of how student achievement effects of More at Four vary with demographics provide evidence of broad benefits for all demographic groups. Student achievement effects are stronger for children whose mother had less than a high school degree. But significant student achievement effects also occur for children of white non-Hispanic mothers with a high school degree or more.

Another way to see evidence for spillover effects is to look at the size of the earnings effects for all children even though only 20% of all children participated in More at Four programs. The 2.9% earnings boost is averaged over all 4-year-olds.  Because only 20% of all children participated, the percentage effects for those children would have to be at least 5 times as great. Furthermore, these at-risk children would tend to have lower lifetime earnings. Assume, based on results from Tulsa, that at-risk children in pre-K might tend to have only 68% of the lifetime earnings of the average person. Then if the More at Four numbers were solely due to effects on children participating in pre-K, the percentage effects on these children’s future earnings would have to exceed 21%.  (21.3%=2.9% times 5, divided by 0.68).  This seems much higher than we would expect from even a high-quality full-day pre-K program for one year.  For example, results in Tulsa imply that a full-day program for at-risk children might boost adult earnings by 9 to 10%.

The implication is that the remainder of the earnings boost is not due to effects on participating children, but rather are spillover benefits for other children. Where might these spillover benefits come from? As Ladd, Muschkin, and Dodge mention, spillover benefits might occur because “once children are in elementary school, all children might benefit from being in classrooms with higher proportions of children who come to school ready to learn and with less need for remediation”.  Children may learn from each other due to peer effects. With less need for remediation for participants in More at Four, teachers have more time to devote to learning for all children, including non-participants. Finally, if the pre-K program also improves behavior, fewer classroom disruptions will allow for increases in all students’ learning.

What is notable is the size of these spillover benefits. Based on the Tulsa results, the estimates imply that for each dollar increase in future earnings benefits for children participating in More at Four, there is another dollar of future earnings benefits for non-participating children. Ladd, Muschkin, and Dodge reach a similar conclusion based directly on the achievement test results, that estimated effects of More at Four are clearly over double what one would expect if there were no spillover benefits.

Spillover benefits are important because they help answer the question: why should I pay higher taxes for pre-K for other people’s children? One answer is because such tax financing is in my enlightened self-interest: it will benefit my child’s future, and the overall economic future of my community.

Given these results, it is unfortunate that the Smart Start program and North Carolina’s pre-K program have faced budget cuts in recent years.  Such budget cuts save funds for taxpayers in the short-run. However, in the long-run, investing in early childhood programs not only benefits participating children, but boosts overall education achievement for all students and boosts the entire state and national economy.

About timbartik

Tim Bartik is a senior economist at the Upjohn Institute for Employment Research, a non-profit and non-partisan research organization in Kalamazoo, Michigan. His research specializes in state and local economic development policies and local labor markets.
This entry was posted in Early childhood programs, Economic development. Bookmark the permalink.

One Response to Published Duke study of North Carolina early childhood programs finds good evidence for spillover benefits for overall student achievement

  1. Phil Gordon says:

    It is a shame that such a well documented successful program is facing cuts. Pay a little early in a child’s life or pay a lot over a lifetime.

Comments are closed.