Estimating the Effect of Higher Education in Germany on Monetary and Non-monetary Outcomes Using the Marginal Treatment Effect

我们使用NEPS(国家教育小组研究)的数据来分析1958-1990年间西德大学教育对2010年货币和非货币结果的长期影响。结果是，一方面，收入和劳动力市场的参与度，另一方面，健康满意度和认知能力(以数学素养、阅读能力和阅读速度衡量)。我们在以下方面对文献做出了贡献：据我们所知，在德国还没有关于大学学习对收入和健康的因果影响的研究。此外，即使在国际文献中，也没有关于认知能力影响的研究。为了得出政策建议，我们使用了两种方法：第一，我们使用概率的外生变异来进行大学研究，以便做出因果声明，而不仅仅是相关性。其次，我们采用了文献中最近的一项创新(边际处理效应)，它允许推导出比传统因果方法(特别是工具变量模型)更大程度上可推广的陈述。这一方法使人们能够估计大学教育对全体人口的平均影响，以及对以前没有实施的潜在政策干预措施的估计。在延伸中，我们的目标是确定允许对大学学习进行成本效益分析的因素。通常，评估研究分析干预措施的总盈余--例如，大学学习是否增加收入的问题。利用文献中的另一项最新创新，我们的目标是确定净盈余，即总盈余减去(通常是主观的)个人学习成本。这对政策制定者来说很重要，因为个人根据预期的净盈余而不是总盈余来决定是否上大学。参考上面的例子，这意味着：即使大学学习增加了收入，如果成本高于收入增长，也不会有回报。

We use NEPS (National Educational Panel Study) data to analyze the long-run effects of university education in West Germany between 1958 and 1990 on monetary and non-monetary outcomes in 2010. The outcomes are, on the one hand, income and labor market participation, and, on the other hand, health satisfaction and cognitive abilities (measured as mathematical literacy, reading competence, and reading speed). We contribute to the literature in the following dimensions: to the best of our knowledge, there are no studies on the causal effect of university studies on income and health in Germany. Moreover, there are no studies on the effect of cognitive abilities even in the international literature. In order to derive policy recommendations, we use two methodological building blocks: first, we use exogenous variation in the probability to take up university studies in order to make causal statements instead of mere correlations. Second, we employ a recent innovation in the literature (the marginal treatment effect) that allows to derive statements that can be generalized to a larger extent than those from conventional causal methods (in particular, instrumental variable models). This approach allows to estimate average effects of university education for the entire population as well as estimations of potential policy interventions that have not been implemented before. In an extension, we aim at identifying factors that allow to perform a cost-benefit analysis of university studies. Usually, evaluation studies analyze the gross surplus of interventions - for example the question whether university studies increase income. Using another recent innovation in the literature, we aim at identifying the net surplus which is the gross surplus minus the (usually subjective) individual costs of studying. This is important for policymakers because individuals decide about university studies based on the expected net surplus and not on the gross surplus. Referring to the example above, this means: even if university studies increase income, they do not pay off if the costs are higher than the income increase.