Quasi-Experiments in Health and Environmental Economics

本计画利用准实验所产生的外生变异，分离出卫生经济学中的一些重要关系。 以下两个广泛的问题进行审查：什么是收入对死亡率的影响？跨越许多学科的大量文献已经确定，收入较低的人健康状况较差，死亡率较高。然而，很难将收入作为这种关系的因果因素。 该项目利用社会保障中经常被称为“福利缺口”的东西作为老年人收入变化的一个外生来源。1977年，联邦政府担心福利金的快速增长，改变了新受益人的福利计算方式，大幅减少了1916年12月31日以后出生的受益人的福利金。由于这些变化，两个生活史相同但出生日期不同的人将获得非常不同的退休收入。 该项目计划比较1916年第四季度和1917年第一季度出生的人的死亡率。如果收入确实会影响死亡率，那么那些每月福利较低的人的死亡率应该会上升。缺口实验是一个很好的机会来研究收入/死亡率之间的联系，原因有四。首先，福利缺口减少了大量的月付款。第二，受影响群体的死亡率相对较高，因此更容易发现收入对死亡率的影响（如果存在的话）。第三，老年人的收入经常被联邦政府改变，如果收入对死亡率有因果关系，可能会产生重要的政策影响。第四，对于狭义的群组，缺口福利似乎接近随机分配，因为在缺口之前和之后出生的工人在立法建立福利差异之前具有相同的观察特征。产前护理的好处是什么？虽然很容易确定产前检查次数更多的妇女有更好的生育结果，但很难将所有这些差异归因于产前检查。那些本来可以顺利怀孕的妇女（例如已婚、年龄较大、受教育程度较高的母亲）也有最高的产前护理使用率。由于潜在的遗漏变量偏差，构建一个令人信服的测试，可以伪造或确认护理和结果之间的联系已被证明是困难的。 这项研究计划使用产前护理访问现场变化，以隔离产前护理的好处。 黑人妇女和城市居民严重依赖公共交通前往医疗保健提供者。因此，可以利用获得交通工具方面的冲击来隔离产前护理减少的影响。 在这种情况下，可以使用一些可能的准实验，包括阿勒格尼县（匹兹堡）和费城的公共交通罢工。来自匹兹堡的初步证据表明，在罢工时怀孕的妇女产前护理的次数大大低于其他时间怀孕的妇女。如果产前护理改善了出生结果，那么罢工造成的就诊减少应该会导致更糟糕的出生结果。

This project uses exogenous variation generated by quasi experiments to isolate some important relationships in health economics. The following two broad questions are examined:What is the impact of income on mortality? A large body of literature that spans many disciplines has established that those with lower incomes have poorer health outcomes and higher mortality rates. However it has been difficult isolating income as the causal element in this relationship. This project uses what is frequently termed the "benefits notch" in Social Security as an exogenous source of variation in the income for the elderly. Concerned with rapidly increasing benefit payments, in 1977, the Federal government changed the way benefits were calculated for new beneficiaries, substantially decreasing payments for recipients born after December 31, 1916. As a result of these changes, two people with identical life histories but different birth dates would receive very different retirement incomes. The project plans to compare mortality rates for people born in the last quarter of 1916 and the first quarter of 1917. If income does impact mortality, there should be elevated death rates among those with lower monthly benefits. The notch experiment is an excellent opportunity to examine the income/mortality link for four reasons. First, the benefits notch reduced monthly payments by a substantial amount. Second, mortality rates in the impacted groups are relatively high, making it easier to detect an impact of income on mortality if one exists. Third, the incomes of the elderly are routinely changed by the Federal government and subsequently, there may be important policy implications if income causally impacts mortality. Fourth, for narrowly defined cohorts, the notch benefits appear to be close to randomly assigned in that workers born just before and after the notch had identical observed characteristics prior to the passage of the legislation establishing benefit differences.What are the benefits of prenatal care? While it is easy to establish that women with more prenatal care visits have better birth outcomes, it is hard to attribute all of these differences to prenatal care. Women who would otherwise have an uneventful pregnancy (married, older, more educated mothers, for example) also have the highest prenatal care use. Because of the potential omitted variables bias, constructing a convincing test that can falsify or confirm the link between care and outcomes has proven to be difficult. This research program uses field variation in prenatal care visits to isolate the benefits of prenatal care. Black women and city residents rely heavily on public transportation to get to health care providers. Shocks in the access to transportation can therefore be used as a way to isolate the impact of reduced prenatal care. A number of possible quasi experiments can be used in this context including public transportation strikes in Allegheny County (Pittsburgh) and Philadelphia. Preliminary evidence from Pittsburgh suggests that women pregnant at the time of the strike had substantially lower prenatal care visits than women pregnant at other times. If prenatal care improves birth outcomes, then the reduction in visits generated by strikes should produce worse birth outcomes.