This Voxeu piece argues that as urbanization surges around the world that it is likely to be a good thing for these urbanites to go to the Megacities. There is a well known tragedy of the commons here. Each rural to urban migrant will compare his private benefits and private costs from moving to a Cairo or a Los Angeles but has no incentive to internalize how his migration decision affects any local externalities (either good or bad) at the potential destination. For example, suppose that traffic congestion or air pollution is an increasing convex function of urban population. In this case, small increases in local population can sharply degrade the city's quality of life. Since pollution and congestion are local public bads, millions of people can be affected by such changes in local public bads.
For example; suppose that an extra population growth of 10,000 people in a city with 8 million people causes local air pollution to increase by 2 units. If per-capita sickness increases by 1 day because of this growth and if each person loses $X dollars per day when sick, then an unintended cost of the growth of the megacity (by 10,000 people) is that 8 million*X*2 of income is destroyed due to the costs of local air pollution.
An environmentalist would say that these 10,000 people should be deflected to some second tier city to avoid these costs of growth. Klaus Desmet and Esteban Rossi-Hansberg appear to dismiss such claims.
In my own micro-economic research, I have argued that the "social costs of mega cities" is falling for two reasons. First, consider the following simple air pollution production function in which air pollution is a function of the population level and the income distribution of the city.
Local pollution = b*local population + F(Income)
I have documented that "b" is declining over time. Due to technological innovation, a growing population causes less pollution. We have documented this both in the United States and in China.
Second, employment suburbanization reduces the impact of urban population on commute times. To understand this point, consider a city where everyone works downtown. An enormous bottleneck ensues as everyone must get downtown. In a city with multiple employment centers, there will be less commuting. To see this , suppose that everyone's home is a separate employment center -- then commute times equal zero!
Now, the authors of this piece are not micro-economists. I bet that what really concerns people is the quality of life of the urban poor. Would the urban poor be happier if they live in a mega-city or a second tier smaller city? An economist would say that by revealed preference -- if they choose to live in a mega-city then it must offer the higher overall package of wages and amenities. A critic would counter that such poor people want to move but are too poor to finance this investment. If this is true, then the Gates Foundation could run some pilots where they offer households a small loan to move and we could see what happens next. This proposal bears some resemblance to this recent Brookings' Hamilton Project paper available as the top link here.