Measuring the Long-Term Impacts of Teachers
[A] teacher with higher value-added raises the probability of college attendance, reduces the probability of teen pregnancy, and increases earnings at age 28.
Concern over the quality of U.S. schools has focused attention on the role that teachers play in school outcomes. Efforts to measure teachers' contributions to student progress generally focus on teacher "value-added," a measure of how much having a class with a particular teacher adds to an individual student's test scores. Debates about the usefulness of value-added measures have centered on whether test scores are adequate measures of the kind of learning that is useful later in life, and the extent to which test scores are affected by parental and school efforts to match particular students with particular teachers.
In Measuring the Impacts of Teachers II: Teacher Value-Added and Student Outcomes in Adulthood (NBER Working Paper No. 19424), Raj Chetty, John Friedman, and Jonah Rockoff conclude that good teachers improve student outcomes up to a decade after they graduate from a large urban school district. The authors match more than one million student records with post-graduation earnings and college attendance information, and then explore the effect of being in a particular teacher's class on earnings and college matriculation.
They find that spending a single year in grades 4 through 8 in a classroom taught by a teacher with higher value-added raises the probability of college attendance, reduces the probability of teen pregnancy, and increases earnings at age 28. A year in the classroom with a teacher who has a value-added that is one standard deviation above the mean is associated with an increase in the probability of college attendance of 0.8 percent, from a mean of 37 percent. The probability of teen pregnancy falls by 0.61 percent from a mean of 13.4 percent; the probability of working at age 28 rises by about 0.4 percent, and annual earnings at age 28 rise by $350, or about 1.7 percent. These effects occur even though the estimates suggest that an individual teacher's influence on test scores falls to 25 percent of its initial impact after several years.
The authors conclude that value-added measurements may be helpful in finding the combination of metrics that best identify the teachers who are most successful at improving long-term student outcomes.