Lead And Mortality Summary

Article Review: Econometrics

“Lead and Mortality” by Clay, Troesken, and Haines presents evidence suggesting that leaching of lead from service pipes into water caused higher infant mortality rates, using US city level data from 1900-1920. The study used multiple sources of city level data to gain insight on the effect of lead exposure via water pipes on infant mortality. This paper will examine the internal and external validity of their research.

Internal Validity Internal validity has two components. First, the estimator of the causal effect must be unbiased and consistent. Second, the standard errors of the estimator must be appropriate to conduct a hypothesis test. Threats to internal validity include omitted variable bias, functional

To control for city characteristics, they used IPUM data to include information regarding population quartiles, white or foreign born, women aged 20-40, and regional fixed effects. It was also necessary for them to control for temperature and precipitation, as these are both believed to effect infant mortality. Other variables included were water quality, milk quality, and women’s suffrage. Water quality can both directly effect infant mortality or through the mother’s health. During this time, increased milk quality was believed to be the reason for the decline in infant mortality. Women’s suffrage may have potentially caused omitted variable bias because it is shown that suffrage led to increased public health spending. Since these variables were found to be correlated with other independent variables and the dependent variable, not including them would have caused omitted variable

Many other populations would differ from this regardless of the year. For example, a population that tends to have children at younger ages may differ sufficiently to threaten external validity. Another example could be a population that different percentage of individuals native to the country or foreign born. The setting in the study was large urban cities in the US from 1900 to 1920. For either a different year in the US, or a different location entirely there will likely be differences in setting. Increased or better health care, in later years in the US or in different countries could differ significantly. Regulation regarding water treatment, milk treatment, and water pipe material could likely also cause a significant difference between