Family and school influences in the determination of child outcomes: Evidence from two sources of variation

在人力资本分析中，一个经典的问题涉及到孩子的结果在多大程度上受到学校和家庭环境的影响。一方面，结果可能完全由家庭决定，而好的学校只不过是好孩子的俱乐部。另一方面，学校可能会真正产生附加值，例如，通过为学生提供好的老师或同龄人。此外，假设家长确实认为学校有影响，一个反复出现的问题是，他们是否将学校质量和自己的努力视为替代品？孩子被一所好学校录取的父母可能会对此做出反应，增加或减少自己的努力。很难提供关于这些问题的确凿证据，主要是因为儿童不是随机分配到学校的，更不用说在家庭环境中分配了。这项研究将通过分析一个环境来解决这一挑战，该环境提供了两个独立的来源，可以说是儿童所经历的学校和家庭环境中的外生差异。具体地说，它将研究出生于和1990年之间的罗马尼亚儿童，这一群体面临两种分析上有用的情况。首先，这些学生是在孩子们通过中央系统被分配到高中的时候过渡到中学的，在这种系统下，入学在很大程度上取决于国家(初中到高中)过渡考试的表现。这就产生了这样一种情况，即过渡成绩非常相似的学生最终进入了质量截然不同的学校。其次，这些人的出生按时间顺序包含了罗马尼亚对堕胎和节育的严格禁令的解除。在这一变化之后，出生率突然下降了30%，这意味着在此之前，大约三分之一的儿童是不受欢迎的，因此出生在不利的家庭环境中。这项研究将使用这两个变异源来实施回归间断和简单的差异估计策略。这两种策略将被应用于(现有的)行政来源和(拟议的)调查的数据，解决以下问题：1)上更挑剔的学校是否会影响学生的认知和行为结果？2)调节这种影响的一些机制是什么？3)出生时的不受欢迎如何影响教育和行为结果？4)这些影响背后的机制是什么？5)家庭环境(由于是不想要/想要的孩子)和学校环境(由于进入更多/更少的选择性学校)如何相互作用来影响孩子的结果？更广泛的影响：关于家庭和学校如何影响儿童结果的这个项目的结果将与发展中国家和发达国家的政策相关。调查结果将通过经济学和教育期刊和研讨会，以及通过与世界银行和多个教育部等机构的直接接触来传播。此外，我们计划在初步分析后公布的数据将对更广泛的研究社区有用。特别是，它的认知和非认知成果的混合将允许有意义的跨学科工作，涵盖教育、经济学和社会学。最后，创建具有详细地理信息的基线调查样本将允许跟踪和研究学生的长期社会经济结果。

In analyses of human capital, a classic issue concerns to what extent outcomes for children are influenced by their school vs. their home environment. On the one hand, it may be that outcomes are fully determined at home, and that good schools are little more than clubs for good children. On the other, schools may really produce value added, for example, by providing students with good teachers or peers. Additionally, assuming parents do think schools have an effect, a recurring question concerns whether they view school quality and their own effort as substitutes? Parents whose children are admitted into a good school may react to this by increasing or reducing their own effort. Solid evidence on these issues has been difficult to produce, mainly because children are not randomly allocated to schools, let alone to family environments. This study would address this challenge by analyzing a setting that provides two independent sources of arguably exogenous variation in the school and home environments that children experience. Specifically, it would study Romanian children born between 1989 and 1990, a group which faced two analytically useful circumstances. First, these students transitioned into secondary education at a time when children were allocated to high schools via a centralized system under which admissions substantially depended on performance in a national (middle school to high school) transition exam. This yields situations in which students with very similar transition scores end up in schools of substantially different quality. Second, the births of these cohorts chronologically contain the lifting of the stringent Romanian ban on abortion and birth control. Following this change, birth rates decreased abruptly by 30 percent, implying that prior to it, about one third of children were unwanted and therefore born into adverse family environments. The study would use these two sources of variation to implement a regression discontinuity and a simple difference estimation strategy. These two strategies would be applied to data from (existing) administrative sources and a (proposed) survey, addressing the following questions: 1) Does attending a more selective school affect students cognitive and behavioral outcomes? 2) What are some of the mechanisms that mediate such effects? 3) How does unwantedness at birth affect educational and behavioral outcomes? 4) What are the mechanisms behind these effects? 5) How do family environment (due to being an unwanted/wanted child) and school environment (due to attending more/less selective school) interact to shape the outcomes for children? Broader impact: The findings of this project on how family and school shape child outcomes will be policy relevant in both developing and developed countries. The findings would be disseminated via Economics and Education journals and seminars, as well as through direct contacts with agencies like the World Bank and multiple ministries of education. Additionally, our data, which we plan to make public after our initial analysis, would be useful to the broader research community. In particular, its mix of cognitive and non-cognitive outcomes would allow for meaningful cross-disciplinary work covering education, economics and sociology. Finally, the creation of a baseline survey sample with detailed geographic information would allow to track and study the longer term socioeconomic outcomes of students.