Do Charter Schools Raise Student Earnings? Evidence from Texas

Texas education policy in the 1990s provided the blueprint for Bush-era school reforms. In 1993, the Texas legislature introduced high-stakes testing, followed by the authorization of Texas’ first charter schools—schools that are publicly funded but privately managed—in 1995. When George W. Bush left the Governor’s Mansion for the White House, 25,000 students were enrolled in charters. By 2015, enrollment had risen to nearly 250,000 students.

Whether or not the proliferation of charter schools is good public policy remains an open question. If charter outcomes exceed those of traditional public schools, the growth in charters could be warranted. Previous research shows that No Excuse charters, such as the Houston-founded KIPP charter network, can improve test scores. However, policymakers are more interested in broader outcomes, such as lifetime earnings, health, and human capital attainment.

Employing Texas administrative data, Will Dobbie and Roland Fryer estimated the impact of charter school enrollment on test scores, college attendance, and early-career earnings, as well as the impact of school quality on these measures. On average, they find that charter school attendance does not impact academic outcomes and is associated with a decrease in earnings of $163 per year. The type of charter school matters, however. Consistent with previous research, No Excuse charter schools outperformed district schools in test scores and four-year college attainment; however, they had no impact on earnings. Non-No Excuse charters, on the other hand, had a negative impact on test scores, college attainment, and earnings.

To arrive at these conclusions, Dobbie and Fryer used matching to identify traditional public school students who could be compared with charter school attendees. Other studies use charter lotteries to identify student matches. This is desirable, because it cleverly avoids the problem of “self-selection”. Because Texas charters are only required to maintain lottery records for two years, they generally did not have these records and matching provided the best alternative. For each charter student in their data, they found one or more traditional public school students who went to the same public school in fourth grade in the same year, and who had the same gender and race as the charter attendee. They then controlled for other characteristics using test scores and services received in fourth grade. Some examples of services are free and reduced price lunch and English language support. Once a match or group of matches was found, they assessed differences in earnings and other outcomes.

In a second avenue of inquiry, the researchers shift focus to the school-level and ask how school quality impacts wages. Overall, moving from a school with low test scores or college attainment to a slightly better one had a sizeable impact on earnings and employment. However, moving from a charter school that performed at the state average to a better one had no impact on labor outcomes. Bluntly stated, bad schools hurt students; but at a certain level schools are good enough. The exception came when high schools were rated by graduation rates; higher graduation rates were consistently correlated with higher earnings. This suggests that high school graduation rates could be a better tool than test scores when evaluating schools.

Taken together, Dobbie and Fryer question the odd result that increased human capital does not lead to increased earnings for students who attend high-performing charter schools. Two methodological issues could account for this. First, it is possible that No Excuse charter alumni moved out of state for higher paying jobs more than their peers, which would not appear in Texas administrative data. They tested this by omitting all participants with no income data and also by imputing missing wage data. In both cases, the results held. Second, their results could suffer from geographic limitations. They only track cohorts from five No Excuse charters, with 60 percent of the roughly 1,040 students going to school in Houston. The other school types had locations spread across a variety of cities and towns.

These results may reflect something deeper about No Excuse schools as well. One hypothesis is that their focus on math, reading, and behavioral expectations leads to the neglect of other skills in social studies or foreign languages, which could lead to higher earnings. In any case, the findings suggest that the expansion of charters requires further analysis.

Article source: Dobbie, Will, and Roland Fryer. “Charter Schools and Labor Market Outcomes.” National Bureau of Economic Research Working Papers No. 22502. (2016).

Featured photo: cc/(recep-bg, photo ID: 627870120, from iStock by Getty Images)