A Dynamic Model of Effort Choice in High School Olivier De Groote* June 11, 2020 Abstract I estimate a dynamic model of educational decisions when researchers do not have access to measures of study effort. Students choose the academic level of their program and the probability to perform well. This differs from a pure discrete choice model that assumes performance follows an exogenous law of motion. I investigate high school tracking policies and obtain the following results: (1) encouraging underperforming students to switch to less academic programs reduces grade retention and dropout, (2) the decrease in the number of college graduates is small, and (3) a pure discrete choice model would ignore changes in unobserved study effort and find a large decrease in the number of college graduates. Keywords: high school curriculum, early tracking, dynamic discrete choice, CCP estimation, effort JEL: C61, I26, I28 *Toulouse School of Economics, University of Toulouse Capitole, Toulouse, France. E-mail:
[email protected]. This paper benefited from helpful comments at various stages from Peter Arcidiacono, Estelle Cantillon, Jan De Loecker, Koen Declercq, Thierry Magnac, Arnaud Maurel, Fran¸coisPoinas, Jo Van Biesebroeck, and Frank Verboven and several audiences at KU Leuven and Duke University. It also benefited from discussions during and after presentations at ULi`ege,RUG Groningen, UAB Barcelona, Toulouse School of Economics, University of Cambridge, LMU Munich, McGill University, UNC Chapel Hill, Oslo University, ENSAE-CREST, IWAEE 2017, EALE 2017, ECORES and LEER summer schools 2017, NESG 2018, CEPR Applied IO 2018, ES 2018 European Winter Meeting, SOLE 2019, and the Barcelona GSE summer forum 2019.