Elizabeth U. Cascio and Diane Whitmore Schanzenbach. 2013.The impacts of expanding access to high-quality preschool education. Working Paper 19735. National Bureau of Economic Research

http://www.nber.org/papers/w19735

Abstract:President Obama’s “Preschool for All” initiative calls for dramatic increases in the number of 4 year olds enrolled in public preschool programs and in the quality of these programs nationwide. The proposed program shares many characteristics with the universal preschools that have been offered in Georgia and Oklahoma since the 1990s. This study draws together data from multiple sources to estimate the impacts of these “model” state programs on preschool enrollment and a broad set of family and child outcomes. We find that the state programs have increased the preschool enrollment rates of children from lower- and higher-income families alike. For lower-income families, our findings also suggest that the programs have increased the amount of time mothers and children spend together on activities such as reading, the chances that mothers work, and children’s test performance as late as eighth grade. For higher-income families, however, we find that the programs have shifted children from private to public preschools, resulting in less of an impact on overall enrollment but a reduction in childcare expenses, and have had no positive effect on children’s later test scores.

摘要：奥巴马总统的“全民学前教育”计划呼吁被纳入公办学前教育的4岁儿童人数的大幅增加和全国范围内此类项目质量的大幅提高。这一计划与九十年代在佐治亚州和俄克拉荷马州推行的普遍学前教育具有很多相似的特点。这一研究通过收集多渠道的资料来估计这些“模范”的州级项目所带来的对学前教育招生和大范围的家庭与儿童群体的影响。我们发现州级项目在增加学前教育招生率方面对于低收入和高收入家庭的儿童的影响是相似的。对于低收入家庭，我们的研究还表明这些项目增加了母亲和孩子共同进行阅读等活动的时间，增加了母亲工作的机会，并提高了孩子成长到八年级时的考试成绩。但是对于高收入家庭，我们的研究发现这些项目将儿童从私立学前教育转到了公立学前教育，对总体上招生率的影响更小一些。不过同时导致了更少儿童照管的开销，对儿童之后的考试成绩也没有影响。

数据来源：国家早期教育研究所National Institute for Early Education Research (NIEER)

研究方法：双重差分分析（difference-in-difference）

Conclusion:President Obama’s $75 billion “Preschool for All” initiative calls for dramatic increase sin the number of 4 year olds in public preschool programs and in the quality of these programs across the nation. This proposal shares—and other proposals likely to follow will share—many characteristics with the state-funded preschool programs Georgia and Oklahoma, which both meet high-quality benchmarks and are accessible to all children. We estimate the impacts of these “model” programs on a variety of child and family outcomes using difference-in-difference frameworks. Our findings suggest sharply different impacts for children from across the income distribution, which is not surprising when one recognizes that the impact of attending a high-quality public preschool depends crucially on what the child would have been doing in the absence of the program.

For lower-SES children, we find evidence of increases in math scores that may be sustained through eighth grade. The increases may be working through multiple channels. First, children are likely to switch from not attending preschool to attending a high-quality public preschool when a universal program is introduced. Moreover, we find suggestive evidence that although they spend less time overall in the presence of their mothers, they spend more time actively engaging in activities such as playing and reading with them. We also find suggestive evidence that their mothers become employed. For higher-SES children, we find no positive impacts on student achievement. These children are much less likely to be moved on the extensive margin of preschool enrollment, and instead are more likely to switch from private to public preschool in response to the program. We also find suggestive evidence that some families are able to substantially reduce their spending on private preschool and childcare in response to the program, freeing these resources up for other purposes.

This pattern of results raises the question of whether the proposal design could be altered to obtain the same positive impacts without inducing as much crowd-out. Could a targeted program meet these goals more efficiently? The findings of rapid and complete fadeout in test score effects in recent randomized controlled trials of Head Start (Puma et al., 2012) and Tennessee’s targeted voluntary preschool program (Lipsey et al., 2013b) suggest that targeted programs today might not induce such gains. One possible explanation is that the test score impacts of universal programs rely on peer effects in preschool classrooms. Indeed, universal programs might be “high quality” not because they meet specific quality benchmarks, but rather because of improvements in the classroom environment from the presence of higher-SES children. We cannot rule out this possibility, and we think it is an important question for future research.

That said, while the Georgia and Oklahoma programs share a number of features with the preschool programs proposed under the Obama plan, there are several reasons to be cautious in generalizing our findings. First, and most importantly, the Obama plan gives states discretion over whether and how much to charge middle-class families for access to the public programs. The more that states charge middle-class families, the less substitution away from private programs there is likely to be. Second, while it has been the focus of our analysis, universal preschool is not all that there is to the “Preschool for All” initiative. For example, it also calls for increases in Head Start enrollment among 3 year olds, which affect the success of universal preschool in ways that we have not been able to measure. On the other hand, our suggestion that the programs in Georgia and Oklahoma have had some lasting impacts on the human capital of lower-SES children might have nothing to do with the quality benchmarks emphasized in the Obama plan. We have estimated the reduced-form impacts of these programs, and the mechanisms at work are not completely clear. The available data have also limited our analysis in important ways that might be rectified in future research.

结论：奥巴马总统750亿美元的“全民学前教育”计划呼吁被纳入公办学前教育的4岁儿童人数的大幅增加和全国范围内此类项目质量的大幅提高。正如其他计划一样，这一计划与九十年代在佐治亚州和俄克拉荷马州推行的普遍学前教育具有很多相似的特点，这两个项目都实现了高质量标准并且面向所有儿童。我们用双重差分法的框架来估计这些“模范”项目对各类家庭、儿童影响的结果。我们的发现表明项目对于收入分配不同层级的儿童影响明显不同，这其实并不令人惊讶，因为我们需要意识到项目的影响跟这些孩子不参加高质量的学前教育会取而代之做些什么。

对于社会经济地位低的儿童，我们发现证据表明他们在整个八年级的数学成绩持续提高。分数的增加可能是通过多种渠道。首先，当全民项目被提供时，儿童可能从不接受学前教育转而参加高质量的学前教育。而且，我们发现证据表明虽然他们总体上更少地在母亲陪伴下度过时光，但他们更多地参与到和母亲一起做活动，比如玩耍和阅读。我们也发现证据表明母亲们开始被雇佣。对于社会经济地位高的儿童，我们没有发现学生成就方面的正面影响。这些儿童更不可能推动学前教育的边际招生率，而是更可能从私立学前教育转到公立学前教育。我们也发现这个项目使得一些家庭实际上能够减少他们在学前教育和儿童照看上的开支，是他们能将资源用于其他用途。

这些结果提出了一个疑问：计划涉及能够变动，不将大量的人排除在外，以获得相同的正面作用。一个定向的计划是否能够更有效地实现这些目标？Head Start对照试验中关于考试分数影响迅速并完全消退的发现和Tennessee的指向性学前计划表明今天的定向项目不一定能有上述收获。一个可能的解释是分数效应依赖于学前教育教室的群体效应。事实上，普遍的项目之所以高效，可能并不是因为它们达到了具体的质量标准，而是因为教室环境由于经济地位高的学生的出席而得到的改善。我们不能排除这一可能性，并且我们认为这是未来研究的一个重要问题。

即便如此，当佐治亚州、俄克拉荷马州的项目与奥巴马的学前教育计划有许多共同特点，我们有理由仔细地归纳我们的发现。首先，最重要的是，奥巴马的计划给予各州关于是否向中产阶级家庭进入学前教育系统收费以及收多少费的自主裁量权。向中产阶级收费越多，则对于私人学前教育的项目替代效应越小。第二，尽管全民学前教育是我们研究的中心，它并不是“全民学前教育”计划的全部。例如，计划也要求发起针对3岁儿童的招生，这会通过我们尚未分析的途径影响计划的成功。另一方面，我们针对佐治亚州、俄克拉荷马州的项目中对低经济地位儿童人力资本的持续影响可能和奥巴马计划中强调的质量标准无关。我们已经估计了这些计划削减后的影响，而且其运行机制并不清晰。可获得的数据也限制了我们的研究，这一点在未来的研究中可能被矫正。

（By 苏小舟）