Restoring middle-class opportunities for more Americans

Nick Kristof’s recent column in the New York Times highlighted some of the economic and social challenges facing many working-class Americans. He focuses on an Oregon friend of his who has faced many economic and social challenges due to his upbringing and the lack of sufficient good-paying job opportunities.

As Kristof suggests, the first step to a solution is having some empathy for those with problems. But the next step is coming up with actual solutions that can work in a cost-effective manner, and that have some evidence of success. Kristof briefly mentions early childhood education, other efforts to boost education, and a higher minimum wage. In this blog post, I’ll flesh out some of those solutions a bit and add to the list of possible solutions.

If the problem is the lack of sufficient number of Americans in higher paying jobs, there are logically four ways to solve this problem:

(1) Labor supply solutions (e.g., early childhood education and other education programs) that augment the quality of Americans’ labor supply, with those higher skills helping to lead to more high paying jobs;

(2) Labor demand solutions (see below) that directly intervene with employers to encourage them to create more high paying jobs;

(3) Institutional interventions in labor markets to encourage more higher wage jobs (e.g., minimum wages and unions);

(4) After-market government subsidies (e.g., the earned income tax credit, subsidies for health insurance, child care subsidies) that provide supplements to market income to boost living standards.

I will focus in this blog post on items 1 and 2, labor supply and demand solutions, both in the interests of shortening the blog post and because there is plenty of writing about minimum wages, unions, and government subsidies.

As I have argued in past blog posts, high-quality early childhood programs can play a significant role in reducing income inequality.  Universal pre-K, combined with income-targeted high-quality childcare, can boost the future earnings of middle class children by about 6%, and the future earnings of lower-income children by close to 30%.  Income-targeted childcare will also directly raise the effective incomes of many working-class households, while also helping these families develop more on-the-job skills and go back to school to build long-run skills and earnings.

But other education programs can also help. Making higher education more affordable may boost post-secondary attendance and motivate students to do better in high school; local programs such as the Kalamazoo Promise provide one model.  High school career academies can help students better make a transition to well-paying careers without discouraging post-secondary attendance.  Evidence from the best charter schools suggests that longer school years, small group tutoring, better use of well-designed testing to improve instruction, and high expectations can significantly boost student achievement.

On the labor demand side, customized job training programs and manufacturing extension programs can help small and medium sized businesses to be more productive and competitive, and can thereby encourage these businesses to expand hiring and be able to pay higher wages. In economically distressed regions, wages subsidies for hiring coupled with public service improvements can significantly improve the economic environment and encourage needed job growth.

There are probably many other labor supply and demand interventions that could boost the availability of higher-wage jobs for more Americans. But the above list focuses on programs for which there is some reasonable empirical evidence for effectiveness, which I link to in the above discussion.

Once we have boosted skills and wages through labor supply and demand policies, institutional policies to boost wages such as a higher minimum wage would be more sustainable, because of higher worker skills. And policies to supplement worker wages, such as the Earned Income Tax Credit, would not need to be as large and costly to boost household income to adequate levels, as market wages and employment rates will be higher.

Restoring more middle-class job opportunities is a huge and challenging task. It will not be achieved by one policy, or in an instant. Rather, it requires a sustained commitment, a broad strategy,  and a willingness to experiment.

Posted in Economic development

Pre-K benefits do not depend on anti-crime benefits and small-scale extremely intense programs

A recent Vox article by Libby Nelson made some useful points about pre-K, but also encouraged some misconceptions.

The article pointed out that in many benefit-cost studies of pre-K, for example of the Perry program and the Chicago CPC program, a major share of the estimated benefits is anti-crime benefits. The article pointed out that these anti-crime benefits did not occur for the Abecedarian program. Finally, the article argued that Perry is a small-scale expensive program, and argued that Chicago CPC was an ages 3-9 program, implying that therefore it would be hard to expect high benefits from less intense large-scale pre-K programs.

It is certainly true that a major benefit of some pre-K programs is reduced crime, and this did not occur for the Abecedarian program. It is also hard to argue with the article’s contention that “providing quality pre-K is more difficult than enrolling children in any kind of early childhood education”.

However, many readers may mistakenly think this means that high benefit-cost ratios for pre-K depend upon these anti-crime benefits. As I have argued in my 2011 book, Investing in Kids, my forthcoming 2014 book From Preschool to Prosperity, and in this blog, there are numerous pre-K programs that have benefits much greater than costs based solely on their benefits for adult earnings.

Adult earnings benefits greater than costs are found for the Perry program, the Chicago CPC program, Deming’s study of Head Start, and the Abecedarian program. Based on test score effects, which are a conservative estimator of adult earnings, adult earnings benefits greater than costs are estimated for state and local pre-K programs in Boston, Tulsa, North Carolina, and many other states.

Adult earnings benefits are even greater if we account for spillover effects of some workers’ skills on boosting overall productivity and wages in the economy.

These high adult earnings benefits greater than costs are for state pre-K programs that are large-scale, and that are not as intense as the Perry program. In addition, the Chicago CPC program shows large adult earnings benefits, as well as anti-crime benefits, for the part of the program that was only pre-K at ages 3-4, disregarding the ages 5-9 component of the program.

The article mentioned that the control group in the Abecedarian study had a low crime rate. The article did not mention that this might imply that Abecedarian’s small anti-crime effects might not be typical of pre-K programs targeted at high-poverty neighborhoods. We would expect the control group in such neighborhoods to have higher baseline crime rates, without the pre-K program. Pre-K programs have more potential in such cases for reducing crime.

My economic argument for pre-K in my various writings has not relied on anti-crime effects of pre-K at all, not because I don’t think they exist, but because I’m trying to focus on what pre-K does for earnings. Pre-K has a very high benefit-cost ratio disregarding any anti-crime benefits, and these high benefits can be achieved with large-scale and less intense programs.

Posted in Early childhood programs

Where is the weight of the evidence, and the burden of proof, for targeted vs. universal pre-K?

The Hamilton Project has released a useful e-book that presents evidence on selected anti-poverty policies. This includes some discussion of pre-K programs, by Elizabeth Cascio and Diane Whitmore Schanzenbach.

The Cascio/Schanzenbach chapter argues for expansion of high-quality targeted pre-K.  My own view, as I have stated previously, is in favor of high-quality universal pre-K. What is the evidence for and against each position?

Part of the issue is political. I simply do not think there will ever be sufficient political support for targeted pre-K programs to enable large-scale access of the poor and near-poor to high-quality pre-K. Therefore, from a political perspective, Cascio/Schanzenbach are supporting a policy proposal that will never be fully implemented.

But I also think there are good economic reasons to support universal pre-K over targeted pre-K. Part of the issue is what evidence one finds more convincing.

Cascio/Schanzenbach rely on their own evidence, from a prior paper analyzing test score effects in Georgia and Oklahoma, to argue that universal pre-K mainly benefits the disadvantaged. The problem with this evidence is that it relies on comparisons between what happens in two states and what happens in other states. There are many unobservables that affect test scores in any two states. These unobservables can bias estimates. Furthermore, these unobservables do blow up standard errors when one fully allows for them. As they acknowledge in their own prior paper, estimation procedures that fully allow for the many unobserved “shocks” that affect state test scores find that most of their estimated effects have very big standard errors, so it is hard to reach definitive conclusions.

It is a tempting strategy to use states as “laboratories of democracy”. The estimation problem is that if one only has two “test subjects” (states) in the “treatment group”, estimation errors tend to be very wide.

In contrast, I would point to what I would argue is stronger evidence, from “regression discontinuity” studies of Boston and Tulsa. (Cascio/Schanzenbach have a footnote referring to the Boston study, but they don’t mention the Tulsa study.) This evidence indicates that these universal pre-K programs have effects on kindergarten entry test scores for middle-class children that are 70% (Boston) to 90% (Tulsa) of effects for lower-income children.

In my opinion, regression discontinuity studies provide “silver standard” evidence of effects of pre-K. These studies are NOT random assignment experiments, which would provide stronger, “gold standard” evidence. However, these regression discontinuity studies seem unlikely to be biased by unobservable differences between the “treatment group” (the children who have completed pre-K, and who are at kindergarten entrance) and the “control group” (the children just entering the pre-K program). Both groups of students either are just entering or have recently finished the same pre-K program, so the two groups do not differ in factors causing families to select or not select a pre-K program, or in factors that might affect how program procedures might lead to the programs selecting certain types of families.

The regression discontinuity procedure is essentially to look at the differences between the two groups of students in scores on the same tests, and then to look at how scores vary with age to control for the fact that the treatment group is on average one year older. We expect to see that the scores will go up with age, and if pre-K has an effect, will show a “jump” (a discontinuity) at the age cut-off that separates children who are just a little too young to yet attend kindergarten, and who are entering pre-K, and children a few days older who are just old enough to enter kindergarten, and who participated in pre-K the previous year.  (See my paper with Gormley and Adelstein for more extended discussion of the regression discontinuity model applied to pre-K.)

In addition to this “silver standard” evidence, we have one random assignment study, of Utah, that provides “gold standard evidence” that pre-K has positive effects for middle-class children.

I would regard the Boston/Tulsa, and Utah studies, as providing stronger evidence than comparisons between Georgia/Oklahoma and other states, largely because of the many unobservable variables that can cause test scores in a state to fluctuate. Having said that, is the evidence that pre-K benefits middle-class children as strong as the evidence that pre-K benefits lower-income children? No, it is not, largely because pre-K for lower-income children has been subject to many high-quality studies, both random assignment and others with good comparison groups, which show that pre-K benefits lower-income children. In contrast, there are fewer studies on pre-K services to middle-class children.

Here is where we get into the highly subjective issue of where the burden of proof should be. One position is to say, if there is no strong and overwhelming evidence that pre-K benefits middle-class children, but that there is such evidence for lower-income children, we should only favor targeted services. We shouldn’t favor a universal program unless there is very strong evidence for benefits to middle-class children.

The other position is to say, given that there is very strong evidence of pre-K benefits for lower-income children, and some good evidence for its benefits for middle-class children, it is more likely than not that pre-K programs benefit middle-class children. And one could argue that this is more than enough evidence to move ahead.

Let’s think through the benefits and costs of moving ahead with universal programs. There is some evidence that this will benefit middle-class children, although not “proof”. There is some evidence that this will increase political support for the programs. At the very least such programs provide some help to working-class voters facing various economic issues. And pre-K programs including middle-class children may have more favorable peer effects that will increase program quality for lower-income children. In my view, the benefits of moving ahead with universal programs outweigh the risks.

In contrast, if we just move ahead with targeted pre-K for lower-income children, we may be missing an opportunity to have more significant effects on overall labor force quality by also assisting middle-class children. We lose any opportunity for positive peer effects from having income-mixed classrooms. And we are setting up a program that is doomed to difficulties in eliciting political support, particularly at the state and local level. There are risks involved in supporting targeted programs over universal programs.

Targeted vs. universal pre-K programs is a debate in which we must go beyond a narrow economics debate over what effects are statistically significant, to a debate over the preponderance of the evidence, and debate over what policy approach is best suited for making progress in a real-world political environment.

Posted in Distribution of benefits, Early childhood program design issues, Early childhood programs | 1 Comment

Achievement gaps at kindergarten entry, income inequality, universal pre-K, and more-intensive early childhood education

Milagros Nores and Steve Barnett have written a recently-released report on how kindergarten readiness and preschool enrollment varies by different groups, including for different income groups. What they document is that at kindergarten entrance, children in disadvantaged groups are far behind children from advantaged groups in school readiness, as measured by cognitive test scores. These kindergarten readiness gaps are not offset by pre-K attendance at age 4, as it appears attendance in quality pre-K is higher for children from more advantaged backgrounds.

For example, they find that children from families with incomes below twice the poverty line have kindergarten entry test scores that average between 0.6 and 0.7 standard deviations lower than the test scores of children from families with incomes above twice the poverty line. To put this in terms that might be more intuitive, average kindergarten entry test scores for low-income children are at about the 40th percentile of all children, whereas average kindergarten entry test scores for higher-income children are at about the 65th percentile of all children, a gap of 25 percentiles. On the other hand, they find that while only about 25% of higher –income 4-year olds are enrolled in pre-K programs that are high-quality, this percentage is still lower for low-income children, at around 18%. Quality pre-K enrollment patterns are reinforcing rather than alleviating kindergarten readiness gaps.

In this blog post, I want to relate these kindergarten readiness gaps to the adult earnings distribution, and see what various designs of early childhood education interventions can do to address these achievement and earnings gaps.

Eliminating the kindergarten readiness gap would do a great deal to help the future prospects of children from low-income families, but would not come close to eliminating all earnings gaps. Based on the estimated effects of kindergarten test scores on adult earnings, if we eliminated these kindergarten readiness gaps, by increasing kindergarten entry test scores of lower-income children by 25 percentiles, we would predict that this would increase the adult earnings of these children by about 17%. This is certainly a significant lifetime earnings boost. But it is far short of the amount their expected earnings fall below that of their more affluent peers. Children from low-income families would be expected to have future adult earnings of 55% of children from higher-income families. A 17% earnings boost would increase this ratio of earnings of children from the two groups from 55% to 64%. (64%=55% times 1.17). Thus, eliminating the kindergarten readiness gap in cognitive test scores would only close one-fifth of the adult earnings gap (9% improvement in initial 45% gap).

(Notes on calculations. These calculations of earnings ratios rely on relative earnings of the parents of these children, and estimated intergenerational correlations of earnings. In addition, the effects of raising kindergarten test scores are estimated based on research by Chetty et al. Bartik, Gormley and Adelstein use similar methods, and similar methods are also used in my forthcoming book, From Preschool to Prosperity).

One can view the glass as either surprisingly half-full or surprisingly half-empty. On the one hand, it is amazing that just increasing kindergarten entry test scores by 25 percentiles can have such profound effects on adult earnings, increasing earnings by over one-sixth. This is probably attributable to “skills begetting skills” (Heckman). Students who enter kindergarten with stronger test scores test will tend to learn more in kindergarten, and so on.

On the other hand, why doesn’t eliminating the starting cognitive test score gap fully equalize earnings? For several reasons. First, eliminating the kindergarten entry gap does not eliminate subsequent gaps in the quality of K-12 education, or in access to quality post-secondary education, which will affect adult cognitive skills. Second, there is more to skills affecting earnings than cognitive skills, so eliminating cognitive skills gaps does not necessarily skill gaps in “soft skills”. Third, adult earnings are not just affected by skills, but by access to networks and wealth that can help a person get a better job.

What can early childhood education do to reduce these achievement gaps, and to reduce income inequality? Quite a bit, but early childhood education cannot fully solve either the achievement gap problem or the earnings gap problem. Based on studies in Tulsa and Boston, universal pre-K would help both lower-income children and middle-class children to improve their kindergarten entry test scores, and by similar percentiles. Realistic projections suggest that kindergarten entrance scores might increase due to universal pre-K by 15 percentiles for lower-income children, and for higher-income children by 12 percentiles. This would only slightly lower the test score gap, from 25 percentiles to 22 percentiles. These test score increases would be predicted to significantly boost earnings of both groups. The expected future earnings of children from lower-income families would increase by about 10 percent, and by children from upper-income families by about 5 percent. Although these earnings increases are significant for both groups, it would only be predicted to slightly reduce the earnings gap – the predicted earnings of children from lower-income families would increase from 55% to 58% of the predicted future earnings for children from higher-income families.

However, these projections understate the potential income redistribution from early childhood education for at least two reasons. First, universal pre-K by design is helping all children to improve their future prosperity. It could do more to improve income distribution by not helping middle-class children, but this hardly makes sense if the program is benefitting these children. However, as discussed further below, other early childhood education programs are better designed as targeted programs.

Second, these test score projections may understate the future earnings impact of universal pre-K and other early childhood education. There is considerable evidence that early test score impacts of pre-K programs may tend to understate long-term earnings impacts. For example, this is true for the Perry Preschool Program. (See my forthcoming book for more discussion of this point. In that book, I project that Perry’s early test score impacts would predict an adult earnings impact of 12%, which is over one-third below the estimated impact based on actual adult outcomes of 19%.) This understatement is probably because cognitive test score impacts do not capture the benefits of pre-K programs for improving non-cognitive skills.

Some early childhood education programs, such as the Abecedarian/Educare program of full-time child care from birth to age 5, seems to work far better for lower-income children than for other groups. These programs should therefore be designed as targeted programs for lower-income families.

Estimates suggest that an Abecedarian/Educare program can increase future earnings of former child participants from lower-income families by 26%. (Again, my forthcoming book has more supportive information behind this calculation.)This is a huge increase in future living standards. However, again, we should remember that it only closes a portion of the earnings gap. If Abecedarian/Educare was implemented for all children from lower-income backgrounds, it would increase their expected future earnings from 55% to 69% of the expected future earnings of children from upper-income families (69%=1.26 times 55%). This cuts the future earnings gap by 1/4th. (45% to 31%)

We could again debate whether the glass is half-empty or half-full. On the one hand, early childhood education programs do not “solve” the income distribution problem in that they do not make the income distribution match some utopian scheme for perfect earnings equality. On the other hand, perfect earnings equality is probably unachievable. It is amazing what can be done to reduce earnings inequality with just a few years of high-quality targeted intervention in early childhood.

Early childhood education programs have the great advantage of being an economic intervention that we know how to do, that will both promote greater economic growth and greater economic equity, at the same time. Early childhood education does not achieve utopian economic justice, but what one program can, in the real world?

Posted in Distribution of benefits

What do we know about pre-K peer effects?

A recent opinion piece by David Kirp in the New York Times argued that it makes no sense to put low-income children in income-segregated pre-K programs, as we do in the Head Start program, because of the importance of classroom peer effects. If low-income children learn more if more of their peers are from a variety of backgrounds, then pre-K programs will be more effective in closing achievement gaps if they are income-integrated programs. This does not necessarily require universal free pre-K (one could imagine some sort of voucher program for low-income children, or some sort of sliding scale fees for a universal pre-K program), but it is one argument in favor of real-world universal programs such as the program in Oklahoma, and against common targeted programs in which low-income children are restricted to their own pre-K classrooms.

(David Kirp is a former newspaper editor and current public policy professor at Berkeley who has written two very good books that directly address early childhood education, The Sandbox Investment, and Kids First. His most recent book, Improbable Scholars, looks at an inner-city school district that he argues has achieved success through a variety of policies and practices that include early childhood education. )

Kirp cites one study of Connecticut preschools to support his opinion piece, by Schechter and Bye (2007). To my knowledge, there are four other preschool studies that look at peer effects, by Henry and Rickman (2007), Mashburn et al. (2009),  Justice et al. (2011), and Reid and Ready (2013). What do we find from these studies?

(1)    Only Schechter/Bye and Reid/Ready directly look at peer income effects. The other studies look at effects of peer pre-existing skills on student learning during pre-K. All of these studies show effects of peers in raising learning during pre-K.

(2)    Henry and Rickman’s results suggest that the magnitude of this effect averages about a 20% spillover: if my peers at the beginning of pre-K have 50% higher skills, we would expect my test scores to be 10% higher at the end of pre-K (10% = 20% of 50%), holding all other pre-K characteristics constant.

(3)    The studies have mixed evidence on whether greater integration of pre-K of children with different characteristics will increase overall pre-K effects for the entire population.


Let me elaborate on that last point. Peer effects at least potentially go in both directions: students with stronger initial skills may have peer effects on their initially less-skilled peers, and students with weaker initial skills may have peer effects on their initially more-skilled peers. If these peer effects are completely the same for all types of children, in all types of classrooms, then peer effects would not lead greater income integration or skills integration to increase overall pre-K performance.

For example, suppose we consider two alternatives: completely segregated classrooms by income and skill level, and completely mixed classrooms by income and skill level. If we go from the segregated to the mixed classroom situation, the lower income or lower-skill students benefit from the influence of their higher income or higher-skill peers. But the upper income or higher-skill students might lose the same amount from the influence of their lower income or lower-skill peers.

The case for income and skills integration is that these peer effects are ASYMMETRICAL, that is differ either across different types of students, or at different levels of integration. For example, suppose that lower-income or lower-skill students on average are very influenced by their peers, but upper-income or higher-skill students on average are not so influenced by their peers. This might well be plausible. One could imagine that learning depends upon the richness of language one hears at school, at home, and at play. If middle-class or middle-skill children already have a higher likelihood of having been exposed to such rich language outside of school, perhaps they are less dependent on hearing such language at school. But lower-skill or low-income children might be more dependent, on average, on hearing such language at school.

(A word might be appropriate about the dangers of generalizing about a group. I fully recognize that there is great diversity of children within any group we might define based on income or some test. The peer effect patterns I am referring to might be tendencies for the average child in the group, and may not be at all true of any individual child.)

As another example, peer effects might differ across classrooms.  For example, perhaps an income or skills-integrated classroom is far better in its “peer effects” than a classroom with 100% low-income or initially lower-skilled children, but perhaps the well-integrated classroom is similar in its peer effects to a classroom with 100% higher-income or initially higher-skilled children. This also seems plausible as a hypothesis. In other words, there might be some threshold or tipping effects of different levels of income and skills integration.

When I say the effects from the current research is mixed, what I mean is that only two studies specifically look at this symmetry, and they find different results. Mashburn et al. (2009) find some evidence that peer effects are larger for children with higher initial skills. On the other hand, Justice et al. (2011) find some evidence that peer effects are larger in going from 100% low-skill classrooms to integrated classrooms than they are in going from integrated classrooms to 100% high-skill classrooms.

However, there is other evidence that also bears on this issue. Both Tulsa and Boston run pre-K programs that include some middle-class children as well as low-income children. For both these cities’ pre-K programs, although there are plenty of middle-class children, they are in a minority: 25% of the Tulsa pre-K children are ineligible for a free or reduced price lunch, and 31% of the Boston pre-K children are ineligible for a free or reduced price lunch.

The evidence from both Tulsa and Boston suggests that whatever negative peer effects MIGHT occur for middle-class children from being in a pre-K program that includes a substantial majority of children eligible for a free or reduced price lunch, these peer effects do not prevent middle-class children from gaining substantially from income-integrated pre-K programs. The gains from pre-K for middle class children in test score percentiles in Tulsa are about 90%of the test score gains for lower-income children; the gains from pre-K for middle class children in Boston are about 70% of the test score gains for lower-income children. In both cases, the predicted dollar effect of these pre-K programs on future adult earnings for middle-class children are at least double program costs.

In other words, whatever academic uncertainty there is about the nature of peer effects, universal income-integrated pre-K programs, if run in a high-quality fashion, appear to be able to achieve substantial benefits for both the poor and the middle class.

Posted in Distribution of benefits, Early childhood program design issues

Whitehurst’s latest comments on pre-K

Russ Whitehurst has some more recent comments on pre-K, this time arguing against a more recent study of Georgia pre-K. This more recent study found pre-K effects on cognitive skills which, averaged across all tests used, had an average “effect size” of 0.69. This is quite high.

(“Effect size” is education research jargon for scaling the effects of some policy on test scores by dividing the effect by the “standard deviation” of the test score across students. This is an attempt to control for the arbitrariness of test score metrics by measuring the test score effect relative to how much this particular test score seems to vary in the sample.)

Whitehurst mainly argues against this study’s validity for two reasons, one of which is a weak argument, and the other of which is a stronger argument. First, he argues that that there’s a problem in all regression discontinuity studies because some pre-K graduates inevitably disappear from the sample when they’re followed up on at the beginning of kindergarten. Although this sample attrition could cause bias in program estimates, a bias which could go in either direction, in practice careful studies find that this bias is small. For example, the Boston regression discontinuity study did numerous tests for possible biases and found no sign of them. The Kalamazoo study did some estimates that restricted the sample to only the same children observed prior to pre-K and after pre-K, and found no significant difference in the estimates.

A second and more valid concern is that the Georgia study has much larger sample attrition due to problems in obtaining consent from the families and schools of pre-K graduates entering kindergarten. Furthermore, there are some signs that this differential sample attrition led to the entering kindergarten sample being somewhat more advantaged.  This differential in family consent rates could have led to more advantaged children being over-represented in program graduates, which might bias the study towards over-estimating program effects. I’m sure these issues will be discussed as this report is submitted to academic journals, and is evaluated and re-estimated during the academic refereeing process.

Whitehurst also expresses some doubt about the large size of the estimated effects. The effects are large, although Whitehurst exaggerates the differentials from other research. The average effect size from previous studies in a meta-analysis by Duncan and Magnuson is 0.35, and in a meta-analysis by the Washington State Institute for Public Policy is 0.31.  These average effect sizes tend to be lower for more recent studies, and for Head Start than for state and local pre-K programs.

The regression discontinuity studies tend to get a bit higher effect sizes. For example, average effect sizes for the regression discontinuity study of Boston pre-K was 0.54.

But, as I have discussed previously, and as Whitehurst has alluded to previously, regression discontinuity studies of pre-K are estimating something a little bit different than other pre-K impact studies. Regression discontinuity studies are studying effects of pre-K for program graduates relative to what would have occurred if they had just missed the age cut-off for pre-K entrance and had not attended this subsidized pre-K until a year later. This means that regression discontinuity pre-K studies are in many cases comparing pre-K with no pre-K, as parents are less likely to enroll children in pre-K if they will not be attending kindergarten the next year. In contrast, other pre-K impact studies are measuring the effects of some public pre-K program relative to a comparison group which will be attending kindergarten the next year, and therefore the comparison group is more likely to attend pre-K. The fact that the comparison group is more likely to attend pre-K probably reduces the net impact estimates for these other pre-K studies.

Which type of estimate is more useful? I think they’re both useful. The regression discontinuity results tell us something about the effects of pre-K versus no pre-K. This is useful for comparison with the gross costs of pre-K. The RD estimates are closer to what a labor economist would call “structural estimates” of the effects of pre-K, which can be useful for modeling the effects of other pre-K programs.

On the other hand, other pre-K estimates tell you the effects of this particular pre-K program versus whatever other pre-K programs are currently available in that particular pre-K marketplace. This is useful if the only policy we are considering is whether or not to adopt this particular pre-K program in this particular pre-K market.  In that case, a benefit cost analysis would have to compare the net benefits of this program versus the extra net social costs of substituting this new program for existing programs. In other words, the new program’s costs may be reduced considerably because it may save in costs on existing pre-K programs, which means it doesn’t take as big an effect size for the program to pass a benefit-cost test.

For both of these types of estimates, extrapolating the estimates to some other pre-K program in some other state or local area requires some assumptions. In general, introducing a new high-quality free pre-K program in any particular local area will result in some increases in pre-K enrollment in this program, and some reductions in enrollment in other programs, with the exact pattern depending on the program being introduced and what is currently available in that market.  Neither the RD estimates, nor the estimated effects of some other pre-K program in some other market, will tell you the net benefits of a new pre-K program in a new market without further assumptions about program take-up of the new program versus the old programs, and without some assumptions about the relative quality of the new program versus the old programs.

In sum, I think the Georgia estimates are only suggestive, because of the problem of differential attrition in the treatment and control groups due to survey non-consent. The estimates may be correct, but this would require further analyses to demonstrate that the survey non-consent problem does not significantly bias the estimates.  Because of this problem with survey non-consent, I would currently give this study a grade of “internal validity” (or “research reliability”) of C, although this grade might be moved up by further estimates by the authors to examine this issue.

However, the Georgia estimates are not representative of most of the regression discontinuity studies, which have done further analyses which suggest that the estimates are not biased by problems with attrition.

Whitehurst also updates his analysis of research to downgrade slightly his grade of “internal validity” (intuitively, research reliability)  of the recent Tennessee study, which found quick fade-out of pre-K test score effects in Tennessee, to A- from A.  But he does not note the factors that lead me to give the Tennessee study a grade for “internal validity” of C: specifically, there was differential attrition due to problems of family consent in the control group in this study, and the few estimates that did not suffer from this attrition bias suggest that the Tennessee program may have had greater effects than are found in the main estimates.

In other words, the Tennessee study actually has stronger evidence of biased estimates than is true of this recent Georgia study. However, for the Tennessee study, the bias appears to be leading the pre-K effects to be under-estimated. There certainly is no good reason to give the Tennessee study a higher grade for research reliability than the Georgia study.

Posted in Early childhood programs

The importance of education, and a pre-K experiment to watch

Two articles recently came to my attention that are of considerable relevance to early childhood education.

First, New York Times reporter Eduardo Porter has an article and interview with economist Thomas Piketty on growing economic inequality. Piketty is the author of a new book on inequality that is getting a lot of attention.

One quotation from Piketty in the interview struck me as particularly relevant to early childhood education, and indeed education in general:

“Historically, the main equalizing force — both between and within countries — has been the diffusion of knowledge and skills.”

I think this summarizes what many economists believe about the role of education. But it is important than one of our leading scholars on economic inequality across the world over the last century agrees with that conclusion.

The policy implication is that if one thinks that inequality is one of the leading social issues of our time, it is imperative to go to great lengths to broaden educational opportunities. Early childhood education is one of the most cost-effective ways to do so, although it should be accompanied by other policies as well.

Second, New York Times reporter Kate Taylor had an article reporting on an experiment testing the “Building Blocks” math curriculum in pre-K. (I thank a tweet from the Human Capital Research Collaborative for drawing this article to my attention.)

One point of note in this article is that this particular curriculum is used in Boston’s pre-K program. As noted in a previous blog post, an article by Weiland and Yoshikawa found extremely high test score effects of Boston’s program. I estimate that this program would increase kindergarten readiness among low-income students sufficiently to increase adult earnings by 15%, which is a huge effect for a one-year program.

An important issue is why Boston’s program is so effective. Perhaps this experiment will tell us whether the math curriculum is key. More time will tell.

Posted in Distribution of benefits, Early childhood program design issues