ECIN Replication Package for "Adding Race and Ethnicity to Microeconomic Databases: An Assessment of Alternative Options"

Estimating differences between racial/ethnic groups often requires merging demographic variables from one dataset to variables of interest in another. A common method merges Home Mortgage Disclosure Act data to property databases. One alternative is to acquire this information from voter registration files; another is to predict race with a name-based algorithm. Compared to Census data, which method is more representative varies by location and group. We explore the practical implications of each method by using the matched samples in two empirical applications. Researchers can arrive at different conclusions about racial/ethnic disparities depending on the method selected.

估算不同种族/族裔群体间的差异时，通常需要将某一数据集中的人口统计变量，与另一数据集的目标变量进行合并。一种常用方法是将《住房抵押贷款披露法》（Home Mortgage Disclosure Act, HMDA）数据与房产数据库进行合并。另有两种替代方案：其一为从选民登记档案中获取此类信息，其二为通过基于姓名的算法预测种族。相较于普查数据（Census data），哪种方法的代表性更强会因地区与群体而异。本研究通过在两项实证应用中使用匹配后的样本，探究每种方法的实际应用影响。研究人员基于所选方法的不同，可能会对种族/族裔差异得出截然不同的结论。