Increased Schooling Reduces Hospitalization Later in Life: New Evidence with Optimal Instruments from the United States

TitleIncreased Schooling Reduces Hospitalization Later in Life: New Evidence with Optimal Instruments from the United States
Publication TypeJournal Article
Year of Publication2023
AuthorsYue, D, Ponce, NA, Needleman, J, Ettner, SL, Lleras-Muney, A
JournalAmerican Journal of Health Economics
KeywordsEducation, Hospitalization, Optimal instruments, Plausibly, Sparse model
Abstract

We investigate the causal effect of education on hospitalization. We apply novel techniques to estimate a sparse model that uses the least absolute selection and shrinkage operator (LASSO) regression with a data-driven penalty to construct optimal cross-validated instrumental variables and select a parsimonious set of controls. This method yields consistent and more efficient estimates relative to conventional instrumental variable procedures and overcomes the limitations of previous studies using compulsory schooling laws in the United States. We also use an approach for a valid inference that allows instruments to be only plausibly exogenous. Using the 1992-2016 Health and Retirement Study, our results suggest that an additional year of schooling in early life lowers the likelihood of two-year hospitalizations later in life by 2.6 percentage points (or about 9.5%). This estimate is robust to different model specifications and plausible amounts of imperfect exogeneity and is similar to the local treatment effect among potential compliers.

DOI10.1086/728403
Citation Keydoi:10.1086/728403