It's also likely that the early construction projects were the low hanging fruit - things that had the greatest impact for the lowest cost. As you run out of those things and start to implement the more complex ideas costs are bound to increase. I'm not sure that would be enough to account for a 20* increase though.
One way to test this theory would be to estimate the cost of completing low-hanging-fruit projects from the early twentieth century that were cancelled (e.g. subways in Cincinnati and St. Louis). I doubt you'd find that it's just as cheap to build them now.
