Worldwide, as income per capita rises, the share of the population working in agriculture diminishes. At present, there is little research investigating the mechanisms underlying this phenomenon. In a new paper, I examine the role of human capital investments in the transition off the farm, using household panel data that is nationally representative of rural Mexico. Does improved access to schools reduce the probability that individuals do farm work?
Many new schools were built in rural Mexico after 1990, providing cross-cohort variation in access to schools within communities. I measure the impacts of building primary, lower-secondary (junior high equivalent), and upper-secondary (high school) schools in an individual’s village when at school age on the probability that he or she works in the farm sector in adulthood. If the returns to education are greater in the non-farm sector and building schools in rural areas makes education less costly, then theory predicts individuals will invest in more education and work more in the non-farm sector after schools are built.
I use the Mexico National Rural Household Survey (Spanish acronym, ENHRUM), which contains work histories for more than 4,000 individuals over thirty-one years. The data are nationally representative of rural Mexico. Three survey rounds in 2003, 2008, and 2011 follow up with the same households, and work histories were recorded for all family members from 1980-2010. These include their migration destinations, whether individuals worked in the agricultural or non-agricultural sector, and employment status (wage-earner or self-employed). I also use community level data from the Secretaría de Educación Pública indicating which schools were located in each community from 1990 through 2010.
Diminishing Farm Labour Supply and Rising Education
Analysis of the ENHRUM data shows that the probability of working in agriculture from rural Mexico decreased by .07 percentage points each year on average from 1980 through 2010. Scaling up by a rural working-age population of 16 million people in 2010, this amounts to a decline of 11,200 people per year. During these years, access to education improved in rural Mexico as well. Figure 1 shows the percentage of rural villages with primary, secondary, and upper-secondary schools over time.
As school access improves, children are likely to attend school for more years. Theory predicts that these educational gains, in turn, influence labour force decisions. Figure 2 shows the percentage of working-age individuals who had access to a lower-secondary school in their home village when they were twelve years old (the age when children would begin lower-secondary) along with the percentage who worked in the agricultural and non-agricultural sectors between 1990 and 2010. As the percentage of individuals with a local lower-secondary school rises, the percentage working in the non-agricultural sector also rises, while the percentage in the agricultural sector falls. This observation leads to the question of whether there is a causal relation between access to school and labour sector decisions.
Methods and Findings
I use the panel of work histories to identify the impacts of school supply in an individual’s home village on the probability of working in agriculture as an adult. I regress a dummy variable for an individual working in agriculture each year on lagged dummies for working in agriculture previous years (since someone who worked in agriculture the previous year is more likely to work in agriculture again), a time trend, and dummy variables indicating whether the individual had a primary school in his/her village when he/she was 6 years old, a lower-secondary school when he/she was 12 years old, and an upper-secondary school when he/she was 15 years old. I also control for village fixed effects and state-by-year fixed effects. Since individuals have little to no control over the level of schooling offered in their villages, nor when schools will be built, the building of schools provides exogenous variation in education across cohorts within villages. I control for village fixed effects since communities vary by unobserved variables that correlate with school acquisition and farm work. The results thus only exploit cross-cohort variation within each village. Finally, state-by-year fixed effects control for unobserved state or regional policies that might promote non-agricultural growth and industrialization while also developing school infrastructure. I find that building a lower-secondary school in an individual’s village reduces the probability of working in agriculture by 1.7 percentage points on average. This result is statistically significant at the 5-percentage point level, and it is large, showing that expanding lower-secondary school provision is an important factor reducing the farm labour supply. The results showed no significant impact of building primary or upper-secondary schools on the farm labour supply.
The results indicate that individuals are about 1.7 percentage points less likely to work in agriculture if they had access to a lower-secondary school inside their village when they were 12 years old. The results are robust to inclusion of village fixed effects and state-by-year fixed effects. The reduced probability of working in agriculture that results from improved school supply explains only part of a negative trend of .07 percentage points per year on average between 1980 and 2010. Figure 3 graphs the predicted impact of building secondary schools on the total farm labour supply between 1990 and 2010. When scaled by the rural working-age population of Mexico each year, the impacts of building more advanced schools in rural villages on the farm labour supply are quite large.
The farm labour supply from rural Mexico is on a downward trend and improvements in education are accelerating the transition out of agriculture. As access to secondary schools in rural Mexico improves, fewer children grow up to work in the farm sector. The U.S. and Mexican agricultural industries will have to adapt by switching to less labour-intensive crops, seeking out new sources of labour (presumably from a country and region where education is not on the rise), or by using more mechanized farm practices that require fewer workers. Technological advances in the farm industry will increase the returns to education within farm jobs, thereby reducing the demand for farm labour overall and also helping the farm sector to retain more workers with higher levels of education at higher wages.
J. Edward Taylor, Diane Charlton, and Antonio Yúnez Naude. “The End of Farm Labor Abundance.” Applied Economic Perspectives and Policy. (Winter 2012). 34(4): 587-598.