In practice, it seems to create a black hole where a school that serves a population that isn't good at tests (e.g disabled kids, non English speakers) gets low scores so everyone who could get a good score leaves and then the school does even worse and gets less funding and then the school is failing and the "fix" is to fire the administration, spend money bringing in new staff who don't know the area or the students, and expect that to solve the systemic issues revealed by poor test scores. To avoid this, schools are indeed incentivized to improve - for instance, by directing low performing students to stay home on the day of the test, or by discouraging students with learning difficulties from attending at all.

This effect is visible in urban centers as well.

Public education is supposed to be a leg up into the world for everyone, not the start of the rich getting richer and the poor getting poorer.

That means we are not using the tests results correctly though. Knowing there is a problem is the first step in fixing it - but it doesn't tell you what the next step is.

The idea that competitive pressure improves school performance is one of those simple, obvious, and wrong Econ 101 ideas.

Even if the bad schools can't respond to incentives to improve, allowing students to choose good schools over bad ones naturally increases utility.

No it doesn't. It could easily result in less social utility by promoting self-segregation and the concentration of funding into high achievement schools where the marginal utility of each additional dollar is substantially diminished at the outset.

The reason why kids from poor families do poorly in school is also the reason why those families are _least_ likely to be selective in their choice of schools. These are not independent phenomena.

So unless the only utility function that matters is freedom to choose your school, approaches like vouchers are unlikely to improve things and could easily not only perpetuate but exacerbate achievement gaps and derivative social issues, like crime.

The evidence indicates that voucher systems improve test scores, graduation rates, and other aspects of both public and private schools, not just the latter:

http://educationnext.org/rising-tide/

  Scaled-up voucher programs like those previously advocated
  by Secretary DeVos show the worst effects. There have
  recently been four statewide voucher programs: Florida,
  Indiana, Louisiana, and Ohio. The Florida study is
  inconclusive, and the others show large negative effects. In
  some respects, the Louisiana results are more convincing
  because the results have been corroborated by two different
  sets of researchers and students were assigned to vouchers
  by lottery—the most rigorous way to evaluate vouchers. In
  terms of providing convincing results, the Indiana and
  Ohio programs, are somewhere in between, but these show
  negative results as well.

  Voucher supporters argue that the results have been worse in
  the recent statewide programs because they have been
  “heavily regulated,” by which they mean the requirements
  that students be tested, that these results be made publicly
  available, and that schools must let in any student who is
  eligible for the voucher. The fact that this fairly minimal
  oversight is considered controversial or heavy-handed,
  however, only reinforces that private schools are designed
  to be exclusive and have little interest in external
  accountability.

https://www.brookings.edu/opinions/why-managed-competition-i...

15 years later, http://news.stanford.edu/2017/02/28/vouchers-not-improve-stu...

I wouldn't expect charter schools to actually change much because the evidence is very clear that the most important factor in achievement, overwhelmingly, is the child's home environment.

Do you have evidence to back up your claim that it is "wrong"?