Economic Explanations
Why is attending schools so important for individuals? Economists offer some key explanations. One of them is that education provides individuals with training, expertise, and knowledge, which are considered assets that enhance their productivity and economic value – earnings. In essence, education is an investment in oneself or in the workforce. Economists refer to this concept as ‘human capital’.
Education, in general, is too broad a term to discuss. Let’s use the ‘years of schooling’ instead for clarity. Intuitively, attending school for many years is believed to lead to higher salaries. If this is true, it can not only explain wage variations among individuals at a micro level but also influence income dynamics within countries at a macro level, underscoring its pivotal role in understanding global economic disparities.
An Empirical Example
–Key Takeaways
The economist Esther Duflo utilized the school construction in Indonesia and found that each primary school constructed per 1,000 children led to an average increase of 0.12 to 0.19 years of schooling, as well as a 1.5 to 2.7 percent increase in wages.
–Background of the Policy
Between 1973 and 1978, the Indonesian government embarked on an ambitious educational revitalization plan amid the oil boom. This initiative included the Sekolah Dasar INPERS program studied in this paper, which saw the construction of over 61,000 primary schools, with a budget exceeding $500 million. According to World Bank data from 1990, the INPERS project stands as the fastest-ever expansion of primary education globally, averaging two schools per 1,000 children aged 5 to 14.
–Data Sources
The data utilized in this study is sourced from SUPAS, focusing on male individuals born between 1950 and 1972. Leveraging individual birthplace data, they align SUPAS individual data with regional-level census data and data on the construction of new schools under the INPERS program across different areas.
–An Exquisite Calculation “Difference in Difference”
Economists always emphasize the importance of causality – we want to know the accurately estimated effect of X on Y. Did people in this Indonesia education program spend more years at school? If so, will this increase his or her wages?
Imagine that there are parallel worlds A and B. Jack chose to attend school for 6 years in world A and earned $10,000 a year. In world B, the cloned Jack chose to attend school for 7 years and earned $11,000 a year. Since all the other conditions are the same, the only difference is the number of years Jack spent at school. Then we can say that this $1,000 increase in world B is because he studied one more year. However, it is hard to find the key to this parallel world. And it is illegal to have cloned Jack. The magic power of causal design is that economists try to find similar people in the real world and make comparisons.
Difference in Difference (DID) is the statistical technique used in the paper to get causal effects. It compares changes in outcomes before and after a treatment between an intervention group (which is the line above shown in the figure) and a control group (which is the line below shown in the figure), to estimate the causal effect of the policy change. In this method, we don’t even have to let the control group be totally the same with the people who got treatment – they just have to have a similar trend in outcome Y before the policy change. Since they have the same trends in the pre-period, it is reasonable to believe that even after the policy change, the constant difference in outcome still exists, so we simply deduct that trend and get the intervention effect.
In this paper, the treatment group and control group are cleverly defined: one that received the program (the treatment group) and one that did not (the control group). How much the program affects someone depends on when and where they were born. Indonesian children usually start primary school between 7 and 12 years old. So, if someone was born in 1962 (12 to 17 years old in 1974), they wouldn’t have directly benefited from the program, which started in 1974. But those who went to school between ages 2 and 6 in 1974 would have been fully affected.
As for birth regions, 91.5% of 12-year-old children still reside in their birth region, and consequently attend school locally. All samples were born before the program implementation, eliminating biases such as deliberate migration to certain regions for childbirth post-implementation or delaying childbirth due to the program.
The policy change involves the establishment of the program. The artificially created cut-off point stems from regression of the number of new schools against the number of children, defining regions with positive residuals as high program areas and those with negative residuals as low program areas. In high program areas, there were 2.44 schools per 1,000 children, compared to 1.54 in low program areas, representing a difference of 0.9 schools per 1,000 children. This disparity underscores variations in program intensity across regions.
The foundation of the DID analysis is the idea that, without the program, the educational outcomes of the treatment and control groups would follow similar trends. To test this idea, we can compare individuals aged 12 to 17 in 1974 with those aged 18 to 24. Ideally, both groups wouldn’t be affected by the policy. This matches what we expected, giving initial support for the idea of parallel trends assumption between the two groups.
| Years of educational attainment | Log(wages) | |||||
| High areas | Low areas | Difference | High areas | Low areas | Difference | |
| Aged 2 to 6 in 1974 | 8.49 | 9.76 | -1.27 | 6.61 | 6.73 | -0.12 |
| Aged 12 to 17 in 1974 | 8.02 | 9.40 | -1.38 | 6.87 | 7.02 | -0.15 |
| Difference | 0.47 | 0.36 | 0.12 | -0.26 | -0.29 | 0.03 |
The table above shows detailed data. In areas with fewer new schools (low program areas), both the average level of education and wages are higher than in areas with more new schools (high program areas). Although education levels increased in both types of regions over time, areas with more new schools saw bigger increases (0.47 years compared to 0.36 years). The results of the DID analysis show that a person born in a low program area and aged 2 to 6 in 1974 received an average of 0.12 additional years of education and earned 2.6% more in wages in 1995.
–External Validity
Smart and critical readers may wonder at this point, is it reliable to evaluate the results based on just one country, one policy, and one group of males? Such questioning is very reasonable, and this article also provides classic papers with alternative estimation results in the “Further Reading” section. On average, the studies show that every additional year of education contributes to a 6-18% increase in earnings.
Conclusion
More and more economists are using models and data to confirm the importance of education—Human Capital Accumulation being one of them. They assist governments in evaluating the effectiveness of education programs. World is our lab.
Further Reading
Adhvaryu, Achyuta, Namrata Kala, and Anant Nyshadham. “Returns to On-the-Job Soft Skills Training.” Journal of Political Economy 131 (8): 2165-2208. https://www.journals.uchicago.edu/doi/10.1086/724320
Aryal, Gaurab, Manudeep Bhuller, and Fabian Lange. 2022. “Signaling and Employer Learning with Instruments.” American Economic Review 112 (5): 1669–702. https://www.aeaweb.org/articles?id=10.1257/aer.20200146
Bailey, Martha J., Shuqiao Sun, and Brenden Timpe. 2021. “Prep School for Poor Kids: The Long-Run Impacts of Head Start on Human Capital and Economic Self-Sufficiency.” American Economic Review, 111 (12):3963-4001. https://www.aeaweb.org/articles?id=10.1257/aer.20181801
Ciccone, Antonio, and Giovanni Peri. 2006. “Identifying Human-Capital Externalities: Theory with Applications.” Review of Economic Studies 73 (2): 381–412. https://www.jstor.org/stable/3700644
Duflo, Esther. 2001. “Schooling and Labor Market Consequences of School Construction in Indonesia: Evidence from an Unusual Policy Experiment.” American Economic Review, 91 (4): 795-813. https://www.aeaweb.org/articles?id=10.1257/aer.91.4.795
Sule Alan, Teodora Boneva, Seda Ertac, Ever Failed, Try Again, Succeed Better: Results from a Randomized Educational Intervention on Grit, The Quarterly Journal of Economics, Volume 134, Issue 3, August 2019, Pages 1121–1162. https://doi.org/10.1093/qje/qjz006