"Each school built for every 100 children led to an average increase of 0.12 years of education and 1.5 percent in wages."
NBER Faculty Research Fellow Esther Duflo uses a major shift in education policy in Indonesia to test the premise that spending more money on infrastructure can in fact increase human capital and reduce poverty. In Schooling and Labor Market Consequences of School Construction in Indonesia: Evidence from an Unusual Policy Experiment (NBER Working Paper No. 7860), Duflo concludes that the Indonesian government's 1973 decision to use part of its oil revenues to build more schools led to significant improvements in education and wages for the generations who benefitted from this new policy.
Duflo points out that when one studies the impact of infrastructure on schooling outcomes, it is often difficult to distinguish between the effects of the school themselves and that of the factors which led to the construction of those schools. If children are better educated in regions which invest more in schools, is it because the investment is profitable or because communities where parents and children value education invest more in schools? Conversely, if children are less educated in regions where there is more investment in schools, is it because schools are harmful or because communities or the government invest more where children are lagging behind?
In 1973, the Indonesian government launched a major school construction program in which more than 61,000 primary schools were built over five years. This was the fastest primary school expansion ever undertaken. It was a radical turnaround in the government's thinking about education, since no investment had been made in the late 1960s. The government also decided to build more schools in regions where primary school enrollments were particularly low.
Duflo finds that the policy initiative was effective in increasing education and encouraged a significant proportion of the population to complete more years of primary school. Those who had already left school when the program was launched (those aged 12 or older in 1974) had completed fewer years of education in regions where more schools were built (this reflects the government's policy of building more schools where education level where low to start with). However those who were young enough to attend the newly built schools took advantage of the opportunity: the education levels rose faster over time in regions where more schools were built.
While the quantity of education certainly increased in the Indonesian experiment, the quality could have suffered. To assess this, Duflo looks at the impact of this program on the most direct measure of productivity: wages. In 1995, those who attended the new schools as children were in the labor market. She finds that the gain in the hourly wage of those who were young enough to attend a new school, relative to the older generations, is larger as the number of schools that were built in their region of birth increases. She calculates that each school built for every 100 children led to an average increase of 0.12 years of education and 1.5 percent in wages. Combining the two results suggests that an additional year in school increases one's wage by about 7 percent. Duflo finds no direct evidence of any deterioration in the quality of education. "But even if it has, the effect on wages shows that this decline was not sufficient to offset the impact of the increase in quantity," she concludes.
Perhaps the greatest insight into the Indonesian policy experiment comes when Duflo compares the cost of building schools to the benefits from higher wages. She warns that this type of analysis requires additional assumptions and should be treated cautiously. Nevertheless, she finds that the cost of the program was smaller than the gains from higher wages for those exposed to new schools.
The cost of the program was large from the outset: 61,000 schools were built at a total of $5 billion in 1990 dollars; just for the construction phase between 1973 and 1979, it amounted to 2 percent of Indonesia's GDP in 1973. Still, even though a school construction program takes a long time to generate positive returns because the costs are upfront and benefits are spread over an individual's life, "The evidence [...] suggests," Duflo writes, "that the program was a profitable investment, with an internal rate of return substantially higher than the average interest rate on government debt in Indonesia over the period."
Duflo adds that the benefits might be even larger once additional factors such as health, fertility and child health are taken into account. She suggests that further research into this area might shed light on the broader social impact of increased spending on education infrastructure in the developing world.
-- Anna Bernasek