suzanne wang

Inequality and the Environment

April 19, 2018

The environment and inequality are closely intertwined. In Flint, Michigan, where thousands of residents were exposed to highly dangerous levels of lead, the majority of residents are black and many are low-income. Third grade reading scores in Flint have dropped by a startling 75% in the past year. Anna Aizer, an economist at Brown University, has a recent paper examining exposure to lead and unequal educational outcomes, specifically school test scores and suspensions. I got to see her speak about it this past week— it’s an interesting work of applied economics! She finds that poor people face more environmental hazards that reinforce poverty across generations.

It is well-established that decreased lead levels are correlated with improved child development. The average reading score gap between white and black students have reduced by half since 1971, while black-white lead level differences have reduced by around 70% in the same time frame. As lead levels have decreased, the amount of crime and school suspensions have also decreased.

This correlation does not mean that less lead directly causes better child outcomes. Take the observation that ice cream sales are correlated with a higher number of drownings. But ice cream can’t possibly cause drowning—there is an additional factor present: ice cream sales and drownings are both also correlated with warmer weather. In the case of lead and childhood outcomes, there could also be hidden variables. Lead exposure is correlated with income, race, and a mother’s education. How can we know that it’s lead, and not these other variables, directly impacting test scores and suspensions?

Aizer finds that a unit decrease in average blood lead levels reduces the probability of being significantly below proficient in reading by 3.1 percentage points and reduces the probability of suspension by 1 percentage point. Beyond just correlation, They study demonstrates a causal link, showing that exposure to lead directly results in lower test scores and school suspensions.

The dataset was particularly good: Rhode Island has a relatively widespread lead-testing program—80% of all three-year-olds have at least one measurement—and robust data points such as test scores, suspensions, and free lunch status on the children being studied.

Aizer singled out maternal education as a particularly tricky variable in her research. Maternal education could be like the hidden warm weather culprit in our ice cream example, explaining both lead levels and children’s school outcomes. It is correlated with lead levels, given that the well-educated tend to have higher incomes and live in better housing less contaminated with lead. It also implies advantages like high-quality face time that improve child test scores and behavior. While the correlation is there, Aizer must disentangle maternal education from lead levels and child outcomes in order to prove causation.

Luckily, Rhode Island data has a feature—lead policy implementation data, which Aizer uses in her study on lead and test scores to prove causation. In our ice cream example, this would be as if the government created an ice cream voucher program in the winter to encourage people to buy more ice cream. Instead of looking at ice cream sales, researchers could look at who received the voucher as a proxy for who consumed more ice cream, then link that to drownings. Since winter voucher recipients are not correlated with weather, the weather is safely controlled for in our analysis.

Aizer uses the probability a home was certified at a child’s time of birth in place of the changes in a child’s lead level. In 1997, landlords were required to obtain lead-reduction certificates to rent their properties, leading to a decrease in lead levels. Though lead-reduction certificates are still correlated with a mother’s education, Aizer looks only at low-income neighborhoods, where she assumes maternal education is consistently low—like in the winter when the temperature is consistently cold. Using the lead certificates allowed Aizer show a causal effect of lead levels on test scores.

These results are just the tip of the iceberg. Aizer noted that lead is the only extensively measured toxin, but there are many other possible toxins to examine such as cadmium, mercury, or arsenic. Aizer also emphasized that lead exposure is a far graver issue in developing countries, resulting in intellectual disabilities of 600,000 children each year, according to a 2014 World Health Organization report.

This research suggests that environment forces affect children’s educational outcomes. Unequal distribution leads certain groups to be more exposed to toxins, and this exposure further perpetuates inequality: a dangerous cycle.