Estimating Returns to College Attainment: Comparing Survey and State Administrative Data-Based Estimates.

Area of Science:

• Higher Education Research
• Labor Economics
• Data Science

Background:

• Massive administrative datasets link postsecondary enrollment with earnings.
• State and federal initiatives increasingly use this data for accountability.
• Existing research often relies on single-state data with limited background information.

Purpose of the Study:

• Provide nationally representative, nonexperimental estimates of returns to college degrees.
• Assess limitations of single-state, administrative data for estimating these returns.

Main Methods:

• Utilized National Longitudinal Survey of Youth 1997 data.
• Tested sensitivity of estimated returns using various sample restrictions and covariates.
• Examined effects of including out-of-state earnings to simulate administrative data limitations.

Main Results:

• Failure to control for student ability leads to upward bias in return estimates.
• Restricting samples to only college enrollees understates degree returns.
• These biases may offset each other, suggesting administrative data can approximate true returns.

Conclusions:

• Discusses advantages and disadvantages of survey versus administrative data for estimating college returns.
• Highlights implications for research and policy using single-state administrative databases.