Does inserting private competitors increase public service performance? Evidence from New Jersey's public school system

There is a long lasting debate about the performance effects of inserting private competitors into public service (quasi-)markets. The neoclassical argument goes that increasing competition from the private sector would... [ view full abstract ]

There is a long lasting debate about the performance effects of inserting private competitors into public service (quasi-)markets. The neoclassical argument goes that increasing competition from the private sector would enhance market forces, so that service users would shop around for the best services, thereby sending market signals to service providers. As a result, service providers would try to adjust their services to keep users, as well as attracting new ones. This would eventually lead to increase public service performance.

A prime example here is the case of public education in the United States, which has in past decades experienced a vast growth of independent, privately-managed schools (so-called charter schools). There are several theoretical arguments why competition from charter schools would affect the performance of traditional public schools. First, it has been argued that competition from charter schools would lead public school administrators to initiate programmatic changes to avoid (or respond to) losing students. These changes may translate into better student outcomes (neo-classical hypothesis). Second, it is argued that school administrators actually do not (or only slowly) respond to increasing market pressures from charter schools. This, in turn, would allow charter schools to cream-off the better performing students from traditional public schools, which would negatively affect their school performance (cream-skimming hypothesis). A third line of argument relates to many US charter school’s mission to primarily cater the needs of disadvantaged students. Therefore, as charter schools are inserted into a school district, poor performance students would leave the traditional public school system. As a consequence, the performance of public schools would go up (reverse cream-skimming hypothesis).

We test these different theoretical predictions using a unique dataset of New Jersey’s public schools that ranges from 2002 till 2011, resulting in more than 21,000 school-level observations. We estimate the performance effects of establishing a charter school, changes in its size and the number of students enrolled, using fixed-effects regression models that control for a wide range of possible confounders at both, the school and the district level (including lagged dependent variables, and time/cluster interactions). Our initial analysis indicates that while the establishment of charter schools increases the performance of traditional schools that operate in the same school districts and its neighboring districts, the expansion of established charter schools decreases public school performance.