I thought this was a very interesting comment by Scott Sumner:

I just read that China is building a 350kph train line from Chongqing to Chengdu.  That’s kind of mind-boggling when you consider the rough terrain.  The track will be 66% tunnels and bridges.  Does this make sense?  My heart says yes but my brain says no.  (I think that is what Matt Yglesias is hinting at above.  He seems to favor these projects on environmental grounds, but understands that some may be too fancy for a country like China.)  And it is a critique that has been forcefully made by Yasheng Huang.  I suppose if you wanted to defend these projects you’d make the following argument:

When China becomes rich these project will pass a cost/benefit analysis.  But they will be too expensive to build.  NYC now wishes it had built a better subway system.  But it is too late.  Construction costs are now too high.  China is building rail lines, subways and airports that are totally inappropriate for a country that is much poorer than Mexico.  But they are highly appropriate for a country twice as rich as Mexico, which is where China will be in 30 years.  I still lean toward the Huang perspective, because the argument I just made ignores the opportunity cost of capital invested in these projects, but I think the alternative view is also defensible.

I don’t actually think it’s too late to invest in needed infrastructure in cities like New York, but obviously such investments cost much more now than they would have years ago. As important, the concentration of interests in New York is much higher, making the building of a sufficient constituency for new investments more difficult. And of course, future wealth and development aren’t independent of the infrastructure investment decisions made now.

But all too often growing cities underinvest in infrastructure given current needs, when they ought to be overinvesting. They’ll never again have an opportunity to build what they’ll likely need as cheaply and easily.