A Call for Evidence in the Debate Over School AutonomyBOCCONI STUDENTS ON THE ECONOMICS SEMINARS ORGANIZED BY IGIER. IN THE SECOND ARTICLE OF A SERIES, SAHANA SUBRAMANYAM REPORTS ON THE WORK OF JOSHUA ANGRIST, MIT, AND STEPHEN MACHIN, LSE
Leading international scholars present their cutting-edge research at Bocconi every year, in front of faculty and students. In order to make this work accessible to a larger audience, Bocconi Knowledge publishes the summaries of the scientific and policy seminars organized by the IGIER research center, written by the students participating in the IGIER-BIDSA Visiting Students Initiative.
We all agree that improving schools and student performance is essential for a country’s social and economic development. But how do we do that? Should we give schools more or less autonomy in choosing which teacher to hire, what to teach and how long to be open for? Should schools be private or public? At the IGIER policy seminar “School Reform: Lessons from US and UK Experiences” of 30 October, two prominent economists, Joshua Angrist from MIT and Stephen Machin from LSE, offered their expert perspective on these questions. Here is a brief summary of their findings.
In the US, schools can either be privately or publicly owned. In standard public schools, all decisions are made by the government, leaving the school with no autonomy. Another type of public schools, called “charter schools” are different: they are publicly funded but independent. They autonomously manage funds, set curricula, recruit teachers etc. One such charter middle school, KIPP Academy in Lynn, Massachusetts, is aimed at improving educational outcomes for marginalized students. It follows a “no excuses” model of education, characterized by long school days, strict behavioral codes, selective hiring. It would seem, then, that by comparing the performance of charter schools with that of standard public schools one should be able to assess the role of autonomy, right?
Not quite. You cannot simply compare students’ test scores at KIPP to those at standard public schools because the difference may have nothing to do with school autonomy. For instance, given the effort and discipline that KIPP demands, it likely attracts only highly motivated students. But perhaps these students would have done equally well if they had been in a standard public school. So, differences in test scores need not be only due to differences in autonomy.
To solve this problem, Angrist and his coauthors cleverly use the fact that oversubscribed charter schools admit students based on a random lottery. They figured that if they compare lottery winners to lottery losers they need not worry about differences in motivation. Both groups in fact applied to KIPP and the only difference between them is that, due to luck, some attend KIPP others do not. The results are amazing: enrollment in KIPP Lynn leads to a substantial increase in test scores. So large that, sustained over 2-3 years, could close the score disparities between black and white students!
Another reform in which lotteries can be used is school vouchers. Here, low-income students can choose any eligible private school, and the state will sponsor their fee. If the vouchers are oversubscribed to, they are awarded by lottery. So once again the comparison between lottery winners and losers can tell us the effect of giving students the choice of which school to attend. Critically, here the results are different: students’ test scores actually decrease under the program! While the voucher program grants more autonomy, it does not result in better performance.
Machin presented his work on school Academies in England. In one of the most radical education reforms in the world, during 1997 - 2010 public schools were allowed to convert to an “academy” conditional on having a sponsor. Being an academy meant that the school had far more autonomy. In 2010, the sponsor condition for conversion was dropped, resulting in a rapid increase in the number of academies. Pre-2010 school conversions were among a few, low-performing schools with large minority student populations. In contrast, post-2010 academies were mostly high-performing schools aimed at advantaged students. Unlike charter schools, academies do not have lottery-based admissions. But Machin addresses the selection problem by using the fact that academies are not new schools. He restricts his analysis of test scores to students enrolled in public schools before their academisation. Using this method, he shows that the pre-2010 schools see large test score gains from higher autonomy, while the post-2010 schools do not. The result is very interesting, for the same policy created very different outcomes based on which schools converted and when.
One general lesson from this discussion is the surprisingly positive impact that both KIPP and pre-2010 academies had on minority students and underperforming schools. This finding challenges the often held idea that more school autonomy worsens outcomes for disadvantaged kids relative to the others. The exact opposite seems to be true here. A broader, related, lesson is that different school arrangements have complex effects, and only careful empirical analysis can determine the way to go. With policy debates becoming highly politically polarized, Angrist and Machin advocate for a more evidence-based policy evaluation.
by Sahana Subramanyam