investinginkids

Over the course of several different blog posts, I have been discussing what both logic and evidence suggest for how the effects of preschool vary across different income groups.  Given this discussion, what are plausible estimates of the effects of a universal pre-k program on different income groups? Chapter 8 of my new book, Investing in Kids, provides original simulations that address this question.

As discussed in the book, the universal program being modeled is a half-day school year program for 4-year olds. The program is assumed to be high-quality in that the program has reasonable class sizes, adequate teacher salaries and teacher quality, and a well-designed curriculum that develops both hard and soft skills.  The program is a voluntary program that is provided for free to all four year olds, so it has universal access but not necessarily universal enrollment: 70% of four year olds are assumed to actually be enrolled.

Based on the evidence discussed in previous blog posts, we would expect the dollar effects of preschool on adult earnings of former child participants to be lower for children from higher income families. However, these effects will not be zero.

Effects of expanded high-quality preschool on children from upper income families will be lower for two reasons. First, children from upper-income families would be more likely to already on their own be enrolled in high-quality preschool, even without the expanded program. Second, high-quality preschool is assumed to have smaller dollars effects for children from upper income families because they already (on average) have other advantages and fewer problems to overcome.

I examine how effects of preschool differ across “income quintiles”. The income quintiles are based on ranking all households by household income. Households are then divided into five equal sized groups by their ranking in household income. The bottom income quintile consists of households with less than around $20,000 in income. The second income quintile is households with $20,000 to $39,000 in income. The middle income quintile is households with $39,000 to $62,000 in income. The fourth income quintile is households with $62,000 to $100,000 in household income.  The top income quintile consists of households with more than $100,000 in household income.

The simulations are in part based on some reasonable assumptions about how the earnings effects, measured in dollars,  of preschool for children from the top four income quintiles compare to the earnings effects  of preschool for children from the bottom income quintile. Earnings effects per child for the second lowest income quintile are assumed to be 19% less than earnings effects per child in the bottom income quintile. Earnings effects per child for the middle income quintile are assumed to be 69% less than earnings effects per child in the bottom income quintile.  Earnings effects per child in both the top income quintile and the next highest income quintile are assumed to be 92% less than earnings effects per child in the bottom income quintile.

These assumptions are conservative assumptions for the benefits of preschool for children from middle income and above families. As detailed in a previous post, there is some evidence that benefits of preschool for children from middle income families might be only one-fifth lower than for children from the lowest income families. These simulations assume a more drastic tail-off of benefits, by over two-thirds for middle income families compared to the lowest income families.

Based on these assumptions, a universal pre-k program has highly progressive effects on a state’s distribution of income.  The net benefits for the lowest income quintile are quite high. After allowing for taxes paid, the preschool program boosts the earnings of the lowest income quintile by over 6% of income. For every dollar that the lowest income quintile pays in increased tax costs for the universal pre-k program, the lowest income quintile receives over $25 in higher earnings:  the benefit-cost ratio for the lowest income quintile is over 25 to 1.

But the second lowest income quintile and middle income quintile also gain from universal pre-k. For example, for the middle income quintile, with income between $39,000 and $62,000, the ratio of earnings benefits from universal pre-k, compared to the taxes this group pays for universal pre-k, is around 3 to 1.

Even though the benefits from preschool decline as household income increases, the middle class can still benefit from universal pre-k. The benefits of preschool for lower income groups are so large, that even if benefits decline dramatically for middle income children, the benefit cost ratio of preschool for the middle class can still be favorable.  This crucial point is sometimes missed by critics of universal pre-k.

For the top two income quintiles, the tax costs exceed the earnings benefits from universal pre-k. But the net loss is modest. For example, for the top income quintile, with more than $100,000 in income, for every dollar of taxes paid for the universal pre-k program, this income quintile gets 32 cents in earnings benefits.  But the net loss is only about 0.1% of income, which is one-thousandth of income.

Therefore, universal pre-k provides significant benefits for a majority of income groups. The bottom three quintiles, comprising the 60% of all households with income less than $62,000, all gain from universal pre-k.  The political feasibility of universal pre-k depends on whether these benefits for the majority outweigh the modest net excess of taxes over benefits for the minority of households with higher incomes.

As I mentioned in a previous post, most of the random assignment experiments for the effects of preschool have focused on disadvantaged students. In general, foundations and governments have not financed random assignment experiments on how preschool has affected more advantaged students. As I have explored in previous posts, there is evidence of the effects of preschool on more advantaged students from studies that use other rigorous methods, but I have not yet cited a random assignment experiment.

For evaluating the effects of preschool on more advantaged students, the only random assignment experiment that I am aware of looks at the Brigham Young University Preschool Program. This program serves a group with considerable advantages in education and various social measures. Most of the parents are college graduate or above. Over 96% are two-parent families, and most of the mothers are not working, so there is plenty of parental time available for paying attention to children.  According to the study authors, “The parents’ educational level, the level of the fathers’ occupation, and the level of family income were all above average for the Mountain States region of the country” (Larsen and Robinson, 1989, p. 136).

Early results from this project found that “social competency…was enhanced through preschool participation …although there were no significant group differences in kindergarten test results”. Later follow-up results as of second and third grade found that “Males who had attended preschool scored significantly higher on the reading vocabulary, total reading, spelling, total language, and total battery components of achievement measures than did males who had not attended preschool.”  In contrast, for females, preschool participation had no statistically significant effects on academic achievement in 2^nd and 3^rd grades.  From the paper, it appears that the female preschool participants, compared to the control group of non-participants, did better on nine of the 11 test comparisons, but the differences were not statistically significant. This study is based on a limited sample size, with information reported for 55 males in the preschool group and 35 males in the control group, and for 70 females in the treatment group and 36 females in the control group.

Given these small sample sizes, it takes fairly large preschool effects for such effects to be detected as statistically significant. The mean achievement gains for males on the overall test battery imply a percentile gain in test scores of a little less than 10 percentiles.

Based on Chetty and his colleagues’ results for how 2^nd and 3^rd grade test scores affect adult earnings, this test score gain for males would be expected to increase average annual adult earnings at ages 25 to 27 by a little over $900 or 5%. (These calculations rely on converting the mean gain in normal curve equivalents to a percentile gain, and then use Chetty et al’s results from column 1 of their Appendix Table 5.)

An increase in annual adult earnings of $900 or 5%, even for only a portion of a full working career, would clearly be sufficient to justify even a very expensive pre-k program.  Therefore, based on the BYU study, we can say that there is evidence from a randomized experiment that pre-k can pay off for males from more advantaged backgrounds.

Studies of pre-k in Tulsa, by William Gormley and his colleagues at Georgetown, provide good direct evidence on the relative effects of a state pre-school program on kindergarten readiness for different income groups. Tulsa participates in Oklahoma’s universal pre-k program, which enrolls over 70% of the state’s 4 year-olds.

Gormley’s studies, like many studies of preschool’s effects on kindergarten readiness, rely on what economists call “regression discontinuity” analysis. This method is well-respected by econometricians and other statisticians. Regression discontinuity analysis is regarded as a rigorous approach to evaluation with a good comparison group. It is not as precise as random assignment experimentation, but is usually unlikely to be biased if correctly applied.  The intuitive idea of regression discontinuity analysis in the case of pre-k is that we are comparing entering kindergarten test scores and entering pre-k test scores, on the same test, of children who are of almost the same age, but some of whom just made the age cutoff for attending pre-k the previous year, and others who had to wait a year because they just missed the age cutoff.

Gormley analyses the effects of Tulsa’s pre-k program on three income groups: those children eligible for a free lunch, those children eligible for a reduced price lunch, and those children ineligible for any lunch subsidy. “Free lunch eligibility” means that family income is below 130% of the federal poverty line for a family of that size. “Reduced price lunch eligibility” corresponds to family income between 130% and 185% of the federal poverty line. The most advantaged income group has income above 185% of the federal poverty line.

Gormley finds that Tulsa’s pre-k program has significant effects on test scores at kindergarten readiness for all three groups. In one study, he reports that the absolute test score increase in points tends to be similar for the free lunch and full-price lunch students, but higher for the reduced-price lunch students. In another study, he finds that percentage effects of test scores tend to be somewhat greater for the lower income students. These findings are consistent because the baseline levels of test scores tend to be lower for the lower income students, so a similar absolute test score effect can be a greater percentage effect.

But how would we expect these test score effects to be reflected in later effects on adult earnings?  This has been studied by Harvard economist Chetty and his colleagues. Their focus is on the relationship between end of kindergarten test scores and adult earnings, but it is reasonable to assume that similar relationships would hold between beginning of kindergarten test scores and adult earnings.

If we use Chetty’s results, we find that the absolute test score effects give us a better predictor of dollar effects on adult earnings. Using information from Chetty et al’s results and Gormley et al’s results, I calculate that in Tulsa, the predicted effect of pre-k on adult earnings would be about one-fifth greater in dollars for reduced price lunch students than for free lunch students, and the predicted effect on adult earnings would be about one-fifth less in dollars for full price lunch students than for free lunch students.

The bottom line is this: Gormley’s results imply roughly similar dollar effects on adult earnings for all three income groups. The variance of effects across income groups imply that the “near poor” (the reduced price lunch students) have somewhat greater effects from pre-k than the “poor”, but the non-poor (the full price lunch students) have somewhat smaller effects from pre-k than the poor.  So there is no simple regularity that the poor always benefit more from pre-k than all other groups, rather some middle group may benefit the most.

Of course, Gormley’s results don’t tell us how effects vary within the full price lunch group. Perhaps effects are quite large for students above 185% but below some higher x% of poverty, but then tail off towards zero beyond some income bracket.

Although regression discontinuity estimates are unbiased, we might also want to see effects of random assignment experimental evaluations of preschool for children from advantaged groups. I will consider that topic in a future blog post.

I have been discussing how the economic development benefits of pre-k vary across different income groups (e.g., lower class vs. middle class vs. upper class).  This is a key issue in deciding on the merits of “targeted pre-k”, which would focus on expanding access to high-quality pre-k for more disadvantaged children, versus “universal pre-k”, which would seek to set up a program that is accessible to all children.  This also helps determine the distributional effects across income groups of any proposed expansion of preschool.

In an ideal world, we would have rigorous evidence measuring how the benefits and costs of preschool vary across children from different income groups. However, all the rigorous long-term research studies of pre-k programs focus on disadvantaged children. This is true of the three best long-term studies, the studies of the Perry Preschool program, the Abecedarian program, and the Chicago Child-Parent Center program. This is also true of many of the short-term studies of pre-k’s impact, for example many of the studies done by NIEER.

Some of the NIEER studies do look at states in which there is there are many non-poor children participating. For example, the Great Start Readiness Program in Michigan (formerly the Michigan School Readiness Program) focuses on children who exceed the Head Start income limit, which is set at the poverty line, but who are below 250% (later 300%) of the poverty line, or have at least one of 24 other risk factors (e.g., single parent, teenage parent, diagnosed family problems, low birth weight, and many others).  Evaluations suggest that this program is highly successful in raising kindergarten readiness for its target group.  Therefore, this study suggests that preschool works for the near-poor/working class/lower-middle class, at least for those with some other risk factor.

Other studies of state pre-k programs consider states that are closer to or at universal coverage. For example, the studies of Oklahoma, which has over 70% of its 4-year olds in the state-funded pre-k program, suggest that this program is also highly effective in raising kindergarten readiness.  West Virginia’s program, which enrolls over half of the state’s 4-year olds, and is not targeted by income in most counties, also seems to be effective in raising kindergarten readiness.

These studies of state pre-k programs with broad access suggest that “universal” pre-k programs can on average be effective. However, these studies do not directly look at the effects of preschool on more “advantaged children”. For example, someone could argue that these significant average effects found in Oklahoma and West Virginia could be due to large effects for disadvantaged children that mask zero effects or even negative effects for more advantaged children.

In subsequent blog posts, I will consider more direct rigorous evidence for effects of preschool for children from more advantaged families.

When we talk about creating “universal” access to preschool, versus expanding “targeted” preschool, we largely are asking whether we should just expand preschool for the poor, versus whether we should also expand preschool for the working class and middle class.

Pre-k enrollment rates are already very high for the upper class. For example, for families with more than $100,000 in income, enrollment rates of 4-year olds in preschool are 89%. Enrollment rates are also high for families with between $60,000 and $100,000 in income, for whom preschool enrollment at age 4 tends to be around 80%.

Of course, we don’t know whether all these preschools are high-quality. But we would suspect that many of these families can afford high-quality preschool and do seek it out and purchase it.

Pre-k enrollment rates at age 4 are next highest for the poor. Preschool enrollment rates are a little over 60% for families with less than $20,000 in income. Much of this enrollment would be in Head Start, and in some state pre-k programs.

Pre-k enrollment rates are lower for 4 year olds in families with between $20,000 and $60,000 in income, where enrollment rates tend to be between 50 and 60%. In cases where these families had to pay all of their preschool costs, we would be concerned that the preschool quality might be low because of too few resources to have a reasonable class size and adequate teacher salaries. Of course, some of this enrollment is in state pre-k programs, which in many (but not all) cases are high-quality programs.

The fundamental issue is: should pre-k expansions be focused simply on expanding the access of poor households to high-quality preschool? Or should pre-k expansions also seek to significantly expand access to high-quality preschool for the working class and middle class? It is this group of households, with income in the $20,000 to $60,000 range, that is most affected by the debate over targeted versus universal preschool.

Of course, the debate over universal versus targeted preschool also has considerable symbolic importance. I will come back to that issue in a future blog post.

In this post, and subsequent posts, I will consider how early childhood programs affect the earnings of different income groups. As part of this discussion, I will also consider the politically contentious issue of whether expanded pre-K programs should be targeted at disadvantaged groups, or be more universally accessible. Both of these issues are discussed in chapter 8 of my book Investing in Kids.

In this initial post on this topic, I want to focus on some things that most researchers agree on about the distributional effects of early childhood programs, and the tradeoff between targeted and universal programs.

First, it is generally agreed that early childhood programs, including preschool, probably have higher returns for disadvantaged groups than for middle-income and upper-income groups.  For example, Steve Barnett, codirector of the National Institute for Early Education Research, and an advocate of universal pre-K, admits that “the weight of the evidence seems to indicate that effects [of pre-K] are somewhat smaller for children who are not economically disadvantaged.”

For early childhood programs such as the Nurse Family Partnership, there is strong evidence from randomized experiments that the effects of the program are stronger for disadvantaged families than for other families.

Second, early childhood programs almost certainly have more progressive effects on different income groups than is true for business incentives. As mentioned in a previous post, the dollar effects of business incentives on the earnings of different income groups tend to decline with household income.  Percentage effects of business incentives on the lowest income households are higher, but only because this group has such low baseline average income.  The lowest income group simply has too low wages and skills to take full advantage of the extra employment opportunities provided by general employment growth.

I don’t know of anyone claiming that the dollar effects on adult earnings of early childhood programs will go UP with average household income. Rather, we expect that early childhood programs have dollar effects on earnings that tend to be higher in absolute dollars for lower income households.  With higher absolute dollar effects for lower income households, the percentage effects on lower income households will be considerably greater. This means that early childhood programs will have more progressive effects than business incentives.

Third, the issue of targeted vs. universal early childhood programs is largely an issue only for preschool. There is general agreement that many early childhood services that are more expensive per child than preschool should be targeted at more disadvantaged groups.

As mentioned, the Nurse Family Partnership is designed to provide services to disadvantaged first-time mothers and their children. The research evidence suggests that this targeting makes sense for this program.

Consider programs such as the Abecedarian program, which provided full-time child care and preschool from birth to age five for disadvantaged children.  Such a program’s gross costs over five years per child are almost $80,000. The Abecedarian program is similar to the Educare schools sponsored by the Ounce of Prevention Fund and the Buffett Early Childhood Fund. As far as I know, no one is suggesting that such an expensive and intensive intervention should be contemplated except for disadvantaged children.  Even if this were suggested, a more universal Abecedarian program would be enormously expensive and almost certainly politically impossible.

Therefore, the targeted versus universal early childhood program issue is largely an issue for interventions of more modest cost and intensity, such as a half-day preschool program for 4-year-olds.  The issue is whether the additional access and preschool quality created by universal pre-k, versus a program more targeted on the disadvantaged, is worth the additional costs. This requires an analysis of both economic and political benefits and costs.  This analysis depends in part on how we think that the effects on adult earnings of pre-k programs vary for different income groups. I will discuss these issues in subsequent posts.