Association for Public Policy Analysis & Management

The family income gap in children’s academic achievement is large and growing in the United States (Reardon 2011), raising concerns about intergenerational social mobility and global competitiveness. Yet it remains unclear whether the association between children’s achievement and family income is causal, because family income is endogenous to a host of other, difficult-to-measure parental characteristics, such as parents’ cognitive abilities, habits and values. This lingering question of causality hinders our ability to direct social policy to reduce achievement gaps and to understand the effects of current policies such as income transfer programs, tax reform and other social welfare interventions.

We take advantage of an emerging natural experiment – the extraction of natural gas in the Marcellus Shale geological formation – to study whether quasi-experimental increases in family income affect children’s academic achievement. Before recent technological advances, deep gas was not extractable and had no economic value; now, families who own mineral rights for land located above the Shale can receive payments upon signing a lease and royalties when gas is sold. There are three exogenous sources of variation in expected Marcellus Shale income: the quality of the shale below a property, the historical property rights in the area, and state law on whether extraction is allowed; Pennsylvania (PA) allows drilling, while New York (NY) does not. The combination of these three factors determines which families living above the Marcellus Shale receive income (in some towns, the average family receives an additional $15,000 per year in income after drilling begins); others, including many living in areas where similar drilling occurs, nonetheless experience no change in income.

Using predicted Marcellus Shale income based on these three factors as an instrument, we will conduct difference-in-difference analyses with multiple comparison groups to isolate the causal effect of family income increases on children’s test scores. We will conduct placebo tests using data from New York, where drilling of the Shale is not permitted and no changes in income occur. We have already conducted a variety of specification and falsification checks of our first stage and found it to be robust (Martin and Ananat 2016). Our results will inform both policies that attempt to address disparities in family income and those aimed at improving children’s academic performance.