Recently, Frances Campbell and her colleagues have published a paper that provides new long-term estimates of the effects of the Abecedarian program. These estimates suggest greater long-run benefits of the Abecedarian program than were projected based on previous results at younger ages.
The Abecedarian program was an experiment conducted in the 1970s in North Carolina. This experiment studied a high-quality full-time child care and preschool program, from shortly after birth to age 5, targeted at disadvantaged children. Because the experiment used random assignment to decide which children were in the treatment group versus the control group, we have good reason to expect that the treatment and control group children were on average the same in observed and unobserved characteristics, and that therefore any subsequent differences were due to the program.
The Abecedarian program was an extremely intensive and expensive intervention. Because the program involved full-time child care and early education for a five-year period, the program ended up costing over those five years almost $80,000 per child (that is, about $16,000 per year for five years). It is important to know whether such expensive intervention brings commensurate greater returns. Does going beyond a year or so of half-day preschool for disadvantaged children pay off for those children and society in better outcomes in adulthood and for the economy?
The Abecedarian program is most similar to the current Educare program. Educare centers have been sponsored by the Ounce of Prevention Fund, and the Buffett Early Childhood Fund, along with other foundations.
Up till now, results for the Abecedarian program were only available up to when the former participants turned age 21. Therefore, prior studies required using these age 21 results to project future educational attainment and earnings for Abecedarian participants. The latest paper provides follow-up information through age 30. These later results enable us to do long-term benefit-cost analyses based on more extensive follow-up.
For many readers, this latest academic paper will be behind a paywall (although the price for purchase is only $11.95). The main findings include that as of age 30, the Abecedarian program program had statistically significant and large effects on educational attainment. The point estimates also suggest that the program had large effects on earnings and income, but these effects were not statistically significant at the typically used significance levels. The Abecedarian program had no discernible effect on the percentage of participants with a criminal record, which contrasts dramatically with results from some other early childhood programs, for example with the Perry Preschool program, which had large effects in reducing crime.
The most noteworthy educational attainment effects are that the program raised average years of schooling completed by a little over 1.1 years, and increased the percentage of participants with bachelor degrees from 6% to 23%. These educational attainment effects are greater than one would have projected based on the age 21 results. (For example, based on the age 21 results, in simulations for my book, I had projected that the Abecedarian program would end up increasing the percentage of participants with bachelor degrees by 12%. This later paper suggests that the true effect is to increase the percentage of participants with a BA by 17 %.)
The earnings effects are statistically insignificant by conventional standards. Specifically, there is an 11% chance that the earnings effects could have occurred by chance even if the true earnings effects were zero, which exceeds the customary 5% cutoff for statistical significance. However, the estimated effects on increasing full-time employment rates are statistically significant. And the point estimates are that the program increased annual earnings from a little under $21,000 to a little over $33,000, an increase of over $12,000. This is a large effect on earnings.
Why are such large effects on earnings still statistically insignificant? This occurs because there is a lot of natural variation in earnings, and the Abecedarian study had a small sample size (52 in the treatment group and 49 in the control group in the study as of age 30). Given the natural variation in earnings, the program would have had to have truly huge effects on earnings for these effects to be statistically significant in this small sample. However, given that the program has statistically significant and large effects on both educational attainment and full-time employment, it seems quite plausible that the program does have large positive effects on earnings.
Finally, why does the program have no effects on involvement in crime, in contrast with the Perry Program? This is difficult to explain by program-specific factors, given that the Abecedarian program essentially is the Perry program plus additional child care and preschool, extending the intervention from half-day to full-day for ages 3 and 4, and adding in full-time child care from birth to age 3. It seems more likely that the difference has more to do with differences between the community contexts of the two programs, North Carolina vs. Ypsilanti, Michigan.
How do these latest findings alter a long-term benefit-cost evaluation of the Abecedarian program? Based on results through age 21, in my book Investing in Kids, I estimated that each dollar invested in the Abecedarian program yielded economic development benefits for a state economy of $2.25. These economic development benefits are the increase in earnings per capita for persons who stay in the state, including both effects on former child participants, and effects on their parents.
I redid my calculations using the new, higher estimated effects on educational attainment based on the age 30 survey. These new estimates show an increase of economic development benefits, per dollar spent, from $2.25 to $2.53, or an increase in this benefit-cost ratio of over 10%.
However, the estimated benefits are even larger if one uses the point estimates for earnings effects. The estimated earnings effects of over $12,000 well exceed what one would expect based on an increase in educational attainment of only a little over 1 year. In fact, I calculate that based on the age 30 results for effects on educational attainment, we would expect an annual earnings effect at age 30 of only $4,117 dollars, about one-third of the point estimate of earnings effects in the latest study.
Does this larger earnings effect make sense? It seems consistent with results from the Perry Preschool Project. In the Perry Preschool study, the program had far greater effects on adult employment rates than one would expect based on its effects on educational attainment. (See Table 11 of my 2006 report.) Specifically, Perry increased employment rates at age 40 about 7 times as much as one would expect based on effects on educational attainment.
One explanation of this pattern is that early childhood programs have effects on skills that are not fully captured by educational attainment effects. These additional skills could include soft skills that are important to success in the labor market.
If we assume the point earnings estimates at age 30 are true effects, and assume similar percentage effects on earnings at other ages, the estimated benefits of the Abecedarian program significantly increase. Under these assumptions, for each dollar invested in the Abecedarian program, there would be state economic development benefits of $3.49, a large increase from the previous estimate of $2.25. Of course, these much larger effects are based on earnings effects that are not quite statistically significant, so we should not place too much weight on these results’ precision.
The bottom-line of the age-30 study of the Abecedarian program is that it strengthens our confidence that Abecedarian-style intensive programs can have large benefits. There are real political barriers to targeted programs for disadvantaged children that cost about $80,000 per child. But if these political barriers can be overcome, these intensive programs provide economic payoffs that more than justify their costs.
Thanks for summarizing the powerful impact of this study. At Tomorrow’s Youth Organization, we provide early childhood education to Palestinian refugees in Nablus. Aside from the HighScope Perry Study, are there other longitudinal studies that look at the effects of early childhood education programs among impoverished communities? Or do you know of any studies that look at the effects of early childhood education programs among at-risk populations overseas?
Thank you for your comment, Clare.
Although most of my research has focused on the United States, I know the World Bank has done extensive work on early childhood programs in a variety of countries. You might want to look at the web page for their work, and some of the news and publications linked to there.
Pingback: Benefits of Early Education Persist at Age 30 «