Charter School Market Share And Performance

One of the (many) factors that might help explain -- or at least be associated with -- the wide variation in charter schools’ test-based impacts is market share. That is, the proportion of students that charters serve in a given state or district. There are a few reasons why market share might matter.

For example, charter schools compete for limited resources, including private donations and labor (teachers), and fewer competitors means more resources. In addition, there are a handful of models that seem to get fairly consistent results no matter where they operate, and authorizers who are selective and only allow “proven” operators to open up shop might increase quality (at the expense of quantity). There may be a benefit to very slow, selective expansion (and smaller market share is a symptom of that deliberate approach).

One way to get a sense of whether market share might matter is simply to check the association between measured charter performance and coverage. It might therefore be interesting, albeit exceedingly simple, to use the recently-released CREDO analysis, which provides state-level estimates based on a common analytical approach (though different tests, etc.), for this purpose.

Let’s focus on the math results, since they vary more widely than those for reading, and start with the state-level results. The scatterplot below presents CREDO effect estimates by charter coverage.*

This is far from a tight relationship, but there may be a little something in there. There are exceptions, such as Arizona and Colorado, both of which have weak charter impacts but very high market share (which means they might be worth looking at). Overall, however, states with lower market share tend to exhibit stronger estimated charter effects (though, again, the relationship is weak-to-modest).

Remember, however, that statewide coverage can be misleading. Charter schools are not uniformly distributed across states, and in many cases are concentrated in just a handful of (often urban) districts. For instance, Louisiana charters serve only about 6-7 percent of the state’s students, but most of these schools are located in New Orleans, where coverage is very high. Ideally, we would want to examine these relationships for local markets (e.g., districts).

If we just eyeball the "local landscape," some districts, such as Newark, Boston and New York City, seem to fit the “small sector, stronger impact” mold, whereas others, such as D.C. and New Orleans, do not.

In any case, even highly simplistic analyses such as that above can be useful in uncovering associations that might warrant further examination.

- Matt Di Carlo

* The data for charter school coverage come from the CREDO report. However, I also redid this analysis using data from the National Alliance for Public Charter Schools "Dashboard," which provides state-level estimates, by year, of the number of charters as a proportion of all schools. Using this measure (averaged across the data years in the CREDO report for each state) does not appreciably change the results presented above.


Is it also possible that larger charter market share means more competition for *students*? My experience talking to school leaders (in charter and district schools) is that they're acutely aware of students' ability to opt to go to different schools and the implications that has for the size and nature of their enrollment.

This could, for example, dilute peer effects within charter schools in competitive markets.



Do the regression statistics suggest that the beta is significantly different than zero? What is the t-statistic on the beta?