TL;DR
By analyzing the world’s first implementation of centralized college admissions in early 20th-century Japan, this paper reveals a stark meritocracy-equity tradeoff. While centralization successfully funneled higher-achieving students into elite tracks—boosting the national production of top bureaucrats and high earners by up to 17%—it simultaneously widened the urban-rural gap, as urban students "crowded out" their rural peers from prestigious career paths for decades.
What Problem Does This Paper Address?
Most modern education systems are trending toward centralization (think the Common App or national matching markets), yet we know surprisingly little about the long-run consequences of these shifts. Does a "fairer" meritocratic system actually produce more talented leaders for a nation, or does it simply reshuffle who gets ahead?
The authors tackle the tension between efficiency (selecting the absolute highest scorers) and equity (ensuring geographic and social diversity). In decentralized systems, students often self-select into local schools to avoid the risk of rejection from top-tier institutions. Centralization removes this "risk," but as this paper shows, that change in strategic behavior has profound, multi-generational effects on a country's elite composition.
Methods
The researchers exploit a unique "natural experiment" in Japan between 1900 and 1930, where the government repeatedly oscillated between decentralized and centralized admission systems for its elite "National Higher Schools."
- The Algorithm: The paper identifies this as the first nationwide use of a matching algorithm—a hybrid of the Deferred Acceptance and Immediate Acceptance (Boston) algorithms.
- Identification Strategy: They use a Difference-in-Differences (DiD) approach, comparing the career outcomes of urban-born vs. rural-born individuals across different birth cohorts that were "exposed" to different admission regimes at age 17.
- Econometrics: They control for prefecture-level characteristics (GDP, number of schools) and use various polynomial time trends to ensure the results aren't just reflecting general modernization.
Datasets
The study relies on a massive effort to digitize historical records:
- Short-run Data: Ministry of Education Yearbooks and Student Registers (1900–1930) to track applications and enrollment.
- Long-run "Elite" Data: The Personnel Inquiry Records (PIR) from 1934 and 1939—essentially a "Who’s Who" of Japanese society—containing biographical data on ~80,000 distinguished individuals (executives, politicians, professors).
- Administrative Records: Comprehensive data on all individuals who passed the Higher Civil Service Examinations (HCSE) to track the career trajectories of top bureaucrats.
Key Insights and Ideas
- Centralization as "Insurance": Under the old decentralized system, students could only apply to one school. High-achievers often played it safe. Centralization allowed them to rank a "dream" school first and a "safety" school second, leading to much more aggressive and risk-taking application behavior.
- Breaking Local Monopolies: Centralization integrated the education market. Students began traveling much further from home to attend school, breaking the "local monopoly" rural schools held over their regional talent.
- The "Crowding Out" Effect: Because urban areas (like Tokyo) had better educational infrastructure, their students scored higher on average. When the system became purely meritocratic and centralized, these urban high-achievers took seats in rural schools that were previously occupied by local rural students.
- Productive Efficiency: The paper argues that elite schools actually provide higher "value-added" to high-achieving students. By matching the best students with the best schools, the system produced a net increase in "top-quality" human capital for the nation.
Key Results
- National Talent Boost: Cohorts admitted under centralized systems produced 15–17% more top-ranking government officials (vice-ministers, governors) than decentralized cohorts.
- Economic Impact: There was a 7–9% increase in the number of top 0.05% income earners nationally.
- The Equity Cost: Centralization increased the number of urban-born elites by 10–20%, directly at the expense of rural-born individuals.
- Persistence: These effects were visible nearly 40 years after the admission reform, proving that the design of a single exam at age 17 can dictate the geographic distribution of a country's leadership for half a century.
Limitations and Future Work
- The "Elite" Focus: The study focuses on the top 0.15% of the population. It does not address how centralization affects the "average" student or the broader labor market.
- Mechanism Ambiguity: While the authors suggest a "human capital" explanation (better schools help smart kids more), they acknowledge it is difficult to fully disentangle this from "signaling" (the prestige of the school) or "networking" (connections made at school).
- Historical Context: The results come from a specific era in Japan. The authors suggest that modern researchers should replicate this by linking administrative tax returns with school district data to see if these meritocracy-equity tradeoffs persist in today's more diverse educational landscapes.
Summary updated