School Fees, Seasonality, and Under-investment in Human Capital

In this study, we test the hypothesis that seasonality affects the likelihood of school attendance and completion, with implications for children's learning. To investigate this, we conduct a randomized pilot study varying the timing of school subsidies across 1,000 pupils in rural Uganda. The randomized controlled trial assigns pupils to one of two treatment arms—receiving a subsidy in Term 1 (post-harvest) or Term 2 (hungry season)—or a control group. Subsidies, paid directly to schools, cover student dues, reducing financial barriers to enrollment.

Households first stated whether they preferred the subsidy in Term 1 or Term 2. Their stated preference influenced treatment assignment, with those preferring Term 1 having a 60% probability of receiving the subsidy in Term 1 and 40% in Term 2, and vice versa. To measure the strength of these preferences, caregivers were presented with trade-offs between different subsidy amounts, allowing us to estimate their reservation price—the financial threshold at which they would switch their preferred timing. While nearly all subsidies were paid directly to schools, a small subset of households received cash to ensure incentive compatibility across all choice questions.

2025-02-03

2025-08-22

School enrollment and attendance: (1) School enrollment status per term (binary: enrolled or not enrolled), (2) Days attended per term (from administrative school records and random attendance visits), (3) Dropout rate (binary: dropped out during the academic year or remained enrolled)

Learning and academic performance: (1) Literacy scores (English and Lusoga), (2) Numeracy scores (basic arithmetic skills), (3) Standardized test scores (from Ugandan National Examination Board's Primary Leaving Exam scores and school administrative data), (4) Grade progression (binary: promoted to next class vs. held back)

Caregiver financial decisions: (1) Ability to pay school fees on time (binary: paid on time vs. delayed payment), (2) Total amount of dues paid per term

Child well-being and time use: (1) Household chores and child labor participation (hours per week), (2) Resilience and psychological well-being indicies (measured via self-reported resilience and depression scales), (3) Aspirations and attitudes toward education (self-reported by children and caregivers)

This pilot study employs a randomized controlled trial to assess how the timing of school fee subsidies affects school attendance and learning outcomes. The experimental sample consists of 1,000 pupils from 9 randomly selected government primary schools in Jinja District, Uganda. To construct the sample, official enrollment data from Uganda’s Education Management Information System (EMIS) was used to identify all students enrolled in the third (final) term of Primary 4 and Primary 6 during the 2024 school year. The final sample was selected using a stratified random sampling strategy, with strata defined by grade, gender, and school. The number of treated pupils assigned to each school was proportional to the school's total enrollment, maintaining a 50/50 gender split.

Randomization was conducted at the individual level within each stratum, with students assigned to one of two treatment arms—receiving a school fee subsidy in Term 1 or Term 2—or to a pure control group. Unlike treatment students, students in the pure control group were not surveyed; only administrative data (such as attendance records) will be collected for them. Caregiver preferences for subsidy timing influenced treatment assignment, with those preferring a Term 1 subsidy having a 60% probability of receiving it in Term 1 and 40% in Term 2, and vice versa for those preferring Term 2.

To measure the strength of these preferences, caregivers participated in a preference elicitation exercise, where they faced trade-offs between different subsidy amounts, allowing for the estimation of their reservation price—the financial threshold at which they would switch their preferred term. Additionally, caregivers participated in a similar elicitation exercise to determine their preference between a direct school fee subsidy and an equivalent cash transfer. For the purposes of this pilot study, the randomization was designed to disproportionately allocate subsidies over cash to maximize insights into the effect of fee timing on attendance.

Not available

Random selection of pupils done in office by a computer; randomization of treatment arm done in SurveyCTO at the time of the baseline survey

Individual

No

n/a

1000 pupils

- 500 treatment pupils (split between Term 1 and Term 2 subsidies, based on preferences)
- 500 pure control pupils