The Influence of Early-life Economic Shocks and Genetics on Cognitive Function

Motivation

It is well established that early-life conditions can affect lifelong health, education, and wages (Almond et al., 2017; Baker, 1992; Cunha and Heckman, 2007). Less is known, however, about the extent to which genetic and human capital investments moderate the influence of early-life conditions on later-life outcomes. We exploit state-level variation in exposure to macroeconomic fluctuations during the Great Depression (GD) to examine interactions between early-life conditions, genetic propensity, and school quality on cognitive function after age 50. Previous studies on the impact of the GD on later-life health have found small effects (Cutler et al., 2007; Hale, 2017). However, recent studies have shown evidence that exposure to exogenous shocks may induce heterogeneous treatment effects by genetic disposition (e.g. Schmitz & Conley, 2017, 2015). In the context of an aging population, identifying the degree to which genetic and early-life factors impact accelerated aging may help to reduce long-term healthcare costs.

Data

We use the Health and Retirement Study (HRS), a nationally representative, longitudinal panel study of individuals 50 plus that contains genetic data and comprehensive information about participants’ socioeconomic background and cognitive function. We focus on individuals born between 1924 and 1933, which covers the period directly before and after the economic collapse of 1929 (N~3,000).

To proxy for local economic conditions, restricted data on state of birth from the HRS is linked to state-level employment data from the Census of Manufactures from 1924 to 1933 (employment-to-population ratios). Furthermore, to explore the potential moderating impact of school quality on later-life cognitive function, we link historical state-level data on pupil-to-teacher ratios, average teacher wages, average term length, and average days attended by students in the school year from the Biennial Survey of Education.

Model

To estimate the impact of the GD on cognitive function, we exploit temporal and geographic variation in economic conditions at the state-year level from 1924 to 1933 using the following difference-in-differences specification:

Y_isct = δEMPLOYMENT_st + X_iʹβ + Z_sc + θ_s + η_c + u(_s*c) + λ_t + ε_isct (1)

where Y is the cognitive outcome of individual i, born in state s, in year c, and observed in year t. EMPLOYMENT is defined as the employment-to-population ratio between conception and age five. X contains individual characteristics and Z includes state-level controls interacted with birth year. The terms θ_s and η_c, are state and year of birth fixed effects. u(_s*c) represents state-specific linear time trends. Standard errors are clustered at the state level. δ captures changes in cognition associated with changes in childhood economic conditions.

Moderating impacts of genetic predisposition and school quality are examined as follows:

Y_isct = δEMPLOYMENT_st + ΩEMPLOYMENT_st * PGS_i + βEMPLOYMENT_St * S_st + Xiʹβ + Z_sc + θ_s + η_c + u(_s*c) + λ_t + ε_isct (2)

where PGS_i is an individual’s polygenic score for educational attainment, and S is a composite score of the school quality metrics outlined above. Coefficients Ω and β capture the extent to which genetic and human capital investments moderate early-life shocks from the GD.