On your podcast recently you asked Ed Glaeser for his political economy model to explain why schools in cities are so bad. I think it may just be schools in American cities that are bad rather than schools in cities in general, and the political economy reason why is probably local control over schools.

I am familiar with the situation in England, where outcomes are better in large cities. English children on free school meals (usually because their parents are on welfare) have substantially better exam results and are a lot more likely to go to university in large cities than in the rest of the country, while children not on free school meals do about as well as in large cities or slightly better.

That said, schools in large English cities were bad 20-30 years ago – in 2001 educational outcomes in inner London were the worst in England – and the improvement coincided with major policy change. Starting in 1990, school governance reforms in England have nearly eliminated the powers of local authorities over schools. Most schools are now ‘academies’ entirely independent from local authorities, and local authorities have very little discretion in how they manage schools theoretically under their control. On the other hand, in the US local government makes more of the decisions on education than in any other OECD country: 72% of decisions in the US are local, compared to the OECD average of 3%.

Why is local control bad for cities? I think it comes from the interaction between selection effects, measurement of school quality and local control. It’s genuinely quite hard for parents to assess school quality, and the data that is comparable and easy to use, like test results, combines underlying quality and selection effects. US Cities are surrounded by independently governed suburbs, some of which will always appear to have better schools than the city because of selection effects. This draws the parents that are most motivated by education to move out of the city and to those suburbs, leaving the cities with an electorate less motivated by education as an issue – these parents are probably also better educated. Local governments have no incentives to compete over the capacity to improve the outcomes of poorly performing children, since this could attract more of these students and make their test results look worse.

The problem of a less interested electorate is aggravated by the complexity of school governance. School districts in cities are usually  run by elected school boards, and often too large to be able to compare to other areas, or be held accountable for the performance of any given school. Urban voters have little in the way of partisan cues to help them choose between candidates – elections are either explicitly non-partisan or dominated by Democrats. This makes it easy for powerful interest groups (teachers unions) to dominate school board elections. Suburbs not only have more interested, more educated voters, but have much smaller school districts: a quick look on Google tells me the best school districts in big city suburbs often run fewer than 5 high schools. Suburban voters can easily compare school performance to other suburbs, and hold their boards accountable.

In summary, schools in American cities are bad because they have strong local government control over education, most often through elected school boards. Suburbs in the metro area can select for high-performing students, and parents concerned over education will confuse that with high school performance and move to the suburbs. This leaves the parents least capable of organising to reform schools concentrated in cities, which, compounded by complex governance and the size of urban school districts, leaves parents in cities incapable of overcoming the resistance of interest groups.

That is from Igor Zurimendi.