This is the 11th in this year’s series of posts by PhD students on the job market.
As a Peace Corps Volunteer in rural Mali from 2010-12, I saw that children were less likely to attend school when their household was busy planting or harvesting their crops. During these times, households had to choose between sending their children to the classroom or the field, and sometimes schooling lost. In my job market paper, I investigate the effect of overlap in the school and farming calendars on children's schooling investments.
In many countries in sub-Saharan Africa (SSA), children often assist with household agricultural production, but this can compete with school since the average school calendar takes up over 50% of total days in the year. Country school calendars do not change often and are highly correlated within regions and former colonial status. Comparing school calendars to farm calendars across SSA, I find that countries with a greater percentage of overlapping days in their school and farming calendars also have lower primary school completion rates (Figure 1). For every additional percentage point increase in overlap, the survival rate to the fifth grade of primary school is 2.4 percentage points lower on average (from a baseline of 71.6 percent). While not causal evidence, it does suggest that overlap may be a potential explanation for up to 40 percentage point differences in primary school completion rates across SSA countries.
Figure 1: Primary School Completion and School-Farming Calendar Overlap in SSA
From a theoretical perspective, a child's time allocation to both schooling and farming is bounded by the total days in the school calendar plus the total days in the farm calendar minus the number of overlapping days between them. As such, when the overlap increases (holding the length of each calendar fixed), there is less total time available for schooling and farming, and it is plausible that households may reduce their children’s time allocations to both activities.
A Natural Experiment: Malawi’s School Calendar Change
To test for the causal impact of the overlap of the school calendar with farming, I study a four-month shift to the school calendar in Malawi—one of only six major school calendar changes across SSA from 1997-2019. Implemented to better align with the government's, universities' and neighboring countries' calendars, Malawi's school calendar change provides a plausibly exogenous shock to overlap between school and crop calendars. Figure 2 visualizes this natural experiment, comparing the policy-induced net change in school days per month to the typical farm calendar in Malawi. The figure shows that the policy transferred school days into a period of greater farm labor demand—the rainy-season planting (specifically to November & December)—increasing overlap between the school and typical farming calendar.
Figure 2: Malawi’s School Calendar Change Increased Overlap with Peak Farming
To obtain causal estimates, I use the Malawi Integrated Household Panel Survey to compare highest completed grade level between school-aged youth differentially exposed to the policy shock. Exposure is determined by an individual’s community crop allocation, since, for example, communities cultivating more soybeans (planted in December) experienced a more impactful increase in school-farming calendar overlap than communities cultivating more sesame (planted in January). Therefore, I conceptualize the policy-induced change in school-farming calendar overlap as a novel shift-share variable. The “shift” is a crop-level shock: the policy-induced change in the number of days a crop's calendar overlaps with the school calendar. The identification assumption is that these crop-level overlap shocks are as-good-as-randomly assigned, which is supported by the non-agricultural rationale for the policy, balance tests, and a pretrend test. Crop-level shocks are then weighted by a community's pre-policy “share” of land allotted to producing each crop, which is allowed to be endogenous. I then regress key outcomes on shift-share overlap controlling for the sum of community crop shares and pre-policy outcome measures, among others. To assuage recent concerns about over-rejection in shift-share estimation, I perform a frontier randomization inference procedure and run share-weighted shock-level regressions.
School-Farming Calendar Overlap Slows Grade Advancement
I find that, after four years, a 10-day increase in school calendar overlap during peak farming periods reduces school advancement by 0.34 grades—one lost grade for every three children—with stronger negative impacts for girls and poorer households. These effects are mainly driven by overlap with the labor-intensive sowing period, in which planting for all rainy-season crops is concentrated into roughly a 45-day period. The magnitude of these effects suggests that accommodating farm labor demand should be a top priority in setting school calendars and that overlap between school and farming calendars may indeed play an important role in explaining differences in schooling across SSA.
Additionally, I find that a 10-day increase in school calendar overlap during peak farming periods leads to an 11 percent decline in children doing any household-farm work during the rainy-season sowing and harvest, affirming the prediction that school-farming calendar overlap acts as a time constraint on both activities. Depending on one’s views of household farm-based child labor, this result could be seen as either encouraging or concerning. Assuaging some possible concern, I also find that while affected households hire more farm labor, their farm profits are not negatively affected. This is interesting, as it suggests that the households either places little value on schooling or overestimate their child’s productivity prior to policy.
What School Calendar Minimizes Overlap with the Farm Calendar?
To identify Malawi's overlap-minimizing school calendar, I simulate other plausible post-policy school calendars and estimate their effects on grade advancement. Specific recommendations for Malawi offer broader lessons for other SSA educational policymakers to consider:
1. Set the school calendar to best align school breaks with periods of peak farm labor demand. In Malawi, the overlap-minimizing calendar starts the second week of January as this aligns the end-of-year break with the labor-intensive sowing period in December.
2. Consider allowing for more local adaptation of the school calendar. In Malawi, school-farming calendar overlap improves (falls) by an additional 12% when communities adopt their own overlap-minimizing school calendar rather than the one calendar that minimizes overlap across all communities on average. One idea: let local school boards declare up to two weeks of school holidays that can be made up at the end of the year—like “snow days” here in Michigan.
Taken together, these findings suggest that the school calendar itself can be an effective policy tool for increasing time in school in SSA. Rather than conceptualizing the time trade-off between schooling and child labor as a zero-sum game, policymakers have the ability via the school calendar to alleviate the constraint on total time available to both productive activities. By avoiding overlaps with the farming calendar, the school calendar can be used to increase school participation among those most vulnerable to missing school due to household farming demands.
James Allen IV is a PhD Candidate in Economics and Public Policy at the University of Michigan specializing in development economics, labor economics, and public policy. In addition to his job market paper, he also studies strategies for improving health care delivery in low-income settings.