Working Paper: NBER ID: w31375
Authors: Zachary Szlendak; Richard K. Mansfield
Abstract: Most U.S. high school courses separate classrooms into standard and honors tracks. This paper characterizes the efficiency and distributional impact of changing the share of students enrolling in honors classrooms. Using a sorting model where students choose tracks by course but schools influence the share choosing honors, we show that administrators’ optimal choices of honors track size require knowledge of treatment effect functions capturing the impact of alternative honors enrollment shares on different parts of the student predicted performance distribution. Using administrative data from North Carolina public high schools, we estimate these treatment effect functions by predicted performance quintile. Across various specifications, we find that smaller honors tracks (20%-30% of students) yield moderate performance gains for the top quintile (~.05-.07 test score SDs relative to no tracking) that decline monotonically across quintiles toward zero for the bottom quintile. However, expanding the honors share beyond 30-35% generates further (small) achievement increases only for the middle quintile, while reducing top quintile gains and causing substantial bottom quintile losses. Since many courses feature honors shares above 35% or do not track, we predict that enrolling ~25% of students in honors in each high school course would improve all quintiles’ statewide performance.
Keywords: No keywords provided
JEL Codes: I20; I21; I24
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| smaller honors tracks (20-30% of students) (Y40) | moderate performance gains for the top quintile of students (D29) |
| honors track size beyond 30-35% (C92) | small achievement increases for the middle quintile (D31) |
| honors track size beyond 30-35% (C92) | reduction of gains for the top quintile (F62) |
| honors track size beyond 30-35% (C92) | substantial performance losses for the bottom quintile (D63) |
| enrolling 25% of students in honors across all courses (I24) | enhance overall performance across all quintiles statewide (I24) |