Research for Second Term Human Capital and Economics

Chetty, Friedman, Rockoff (2014)

• Statistical Model theorized that VA is constant

• Measures teacher value-added for a large urban school district in the United States.

  • Data on student test scores, student demographics, and teacher assignments 

  • 2.5 million children, 1989-2009

• They found a lot of variation in teacher quality, as measured by “value-added”

A 1 sd improvement in teacher VA by .14 sd in math and 0.10 sd in English

• Larger variation in math teacher than English teachers

• Second part: found that test score VA is a good proxy for other outcomes we care about

Talent Transfer Initiative

• Paid high-VA teachers to transder to low-preforming schools 

• Implemented across 7 states, 2009-2011

• Finds: The high-VA teachers had no effect on test scores in maths or reading in middle school. They had a small positive effect on test schools in elementary school. 

  • Teachers who high VA didn’t have the same VA at the other schools because of other factors 

Arteaga (2018)

• Holds signal constant, but varies human capital

  • Classes did cause human capital, and the quality of the signal doesn’t change.

Evaluates a reform that reduced the required number of credits to obtain an economic degree by 20% at a university in Colombia

• Potentially affects human capital, but does not affect signal

• Finds that salary declines by 16% for economic majors 

  • Reduced employment probability at high-paying firms 

  • No actual salary reductions with the same firm (entry salaries are fixed) 

• Does not refute the signaling model, but shows that economics courses produce skills that are valued by the labor market. 

Clark and Martorell

• Varying the Signal and holding human capital constant (the range between passing and not passing would have relatively similar human capital)

  • Regression Discontinuity Design (RDD)

• Evaluated the signaling value of a high school diploma in Texas in the mid-1990s

• Found no effect of passing and obtaining a diploma on initial earnings.

• No evidence that high school completion provides a useful signal of productivity

  • Maybe high school is too easy?

  • No asymmetric information? Employers may obtain the information they need in interviews

  • It does not address the signaling value of college or graduate degrees.

  • Does not address the signaling value of high school in a population with low completion rates.

Muralidharan and Sundararaman (2011)

• Found Individual Incentives are efficient, because all methods cost the same

• Teachers’ bonus pay increased linearly with the fraction of students whose test scores increased over the course of a year.

  • Average test score gain must exceed 5% for the teacher to receive a bonus

  • Value-added incentive scheme

• Two treatment arms with teacher incentive pay:

  • Individualized incentive pay: teachers are paid based on the average test score gain of their own students

  • Group incentive pay: teacher paid based on the average test score gain of the entire school 

• Found: Individualized teacher incentive pay produced the highest test score gains 

• Spillover effects in subjects where teachers did not receive performance pay 

• Broad-based gains: weaker and stronger students benefited equally 

Employer Learning Model

• Productivity should be easier to observe for workers employed for many years

  • If education is a valuable signal in hiring, it should matter less for wage as the worker ages. wages

  • The correlation between wages and schooling should decrease with experience. 

  • The correlation between wages and predetermined ability (scores on cognitive tests) increases with experience.

• When employers can’t observe productivity they use education

• Asymmetric information leads to employers having this belief system

• The more experience at a company, the less critical education is, because employers can observe productivity