Real-life voucher programs, as Scott points out, tend to accommodate far too few students to make a difference in the systemic way we educate poor children in this country. But to respond to a point Megan McArdle made in an earlier post on vouchers, let's imagine for a moment that every child in an underperforming urban school district receives a voucher to attend private school. Since there definitely aren't enough existing parochial and private schools to meet the demand, new schools open to take advantage of the government subsidies.
This will still be a drastically unequal system. Why? Because many, if not most, of the new schools will continue to cater exclusively to poor, non-white students. Those schools will suffer from poor reputations (racism and classism are real), less parent volunteer time, less investment from the community, and probably less funding. Megan, one reason to support the federal government providing any service is that greater centralization can reduce inequalities. Many of the inequalities already present in our educational system are the result of state and local funding apparatuses for the schools that rely upon local property taxes. The inequalities are also the result of many wealthier, whiter people drawing boundary lines so that their kids' schools cater only to people like them.
I agree with Megan that it's a problem so many affluent families opt-out of the public schools. But vouchers aren't the solution. We need to decrease the isolation and concentration of already stigmatized groups within our education system. Since there is no evidence that private charters do a better job at educating kids than public schools do, what makes us think crowding poor kids into private schools en masse will fix the problem? Rather, we need to make more public schools into good public schools, so that more parents opt-in. This doesn't have to take decades. Schools can turn around in a year or two under good leadership and with quality teachers and high academic standards.