Based on feedback from the old thread more than a year ago and some new information, I crunched the numbers again for the cost/benefit analysis of improved cycling infrastructure. The numbers are surprising.
First, the costs of the car. Taking the US as a whole, people drove around 2 trillion (2 x 10^12) mi a year. Assuming a modest average 25 mi/gal, that means 80 billion gallons of gas were burnt to power said cars, which generated 776 million tons of carbon dioxide. The carbon dioxide's cost to the environment, if we accept the European Climate Exchange's pricing ($17.50/ton), is a total of $13.58 billion. (I got that number from slate.com's reference about the cash-for-clunkers program.)
The gas itself cost a total of $160 billion dollars, assuming $2/gal (since we can't count the taxes that fund the government to some degree).
Now generously assuming that each car lasts 200,000 mi, so each year 10 million cars would be "spent." Let's assume that the cars cost, including insurance, an average of $20,000 each. That would be $200 billion right there. (I am being generous here.)
So the net cost of cars in the USA per year is roughly $13.58 (carbon dioxide) + $160 (gasoline) + $200 (cars themselves) = $373.58 (all in billions). I find it interesting how money-wise, the carbon effects are a small percentage of the cost of the cars. Using the European carbon price, gas should have a $0.17/gal tax to pay for its negative environmental (carbon) effects.
That being said, the health-care costs attributed to obesity are around $140 billion/yr.
Now, let's talk infrastructure. If an additional 1% (my old 10% figure was shot down as WAY unrealistic) of miles traveled by car can be replaced by bike, and we get a 1% reduction in obesity, the money saved would be $5 billion/yr. The 1% obesity reduction does not have to come from the obese, paradoxically in that improved health in others that cause an equivalent cost reduction works too. With an assumption of a very high cost of $500,000/mi for cycling/walking infrastructure improvements and maintenance (complete the streets $$$), we can have 10,000 mi worth of infrastructure. Divide that among the 100 largest cities in the country, and we'll have 100 mi worth per town per year, which is very good.
Of course, there soon would reach a point where every place is built out and saturated, at which point we go into a strictly maintenance mode which costs much less. And it's not realistic to expect a crash infrastructure building boom, at least without causing massive economic disruption.
I'm assuming that bikes and extra food don't cost anything, which of course isn't true. But I would say that the costs of bikes are around 1% of that of a car, so it's probably negligible in this analysis. Next time opponents complain about the high costs of infrastructure, maybe pointing to an analysis similar to this one would show how it's not really so high.