In reply to economists in support of Keystone XL and my quote "but it seems like the potential environmental costs could be efficiently internalized," Wes comments:

When, in the history of US Politics, has this ever happened? The costs of bad air in California exceed $20 B per year. All of that has been externalized.

First, internalization of negative externalities doesn't mean that there are no more environmental damages. Internalization is realized with the marginal benefits of damages is equal to the marginal cost of damages. The benefit of damages is the producer and consumer surplus for the market goods whose production and consumption causes the damages. The cost of the damages is the lost health, recreation and other amenities.

The Keystone XL analysis should compare the market benefits of the oil with the environmental costs. The market benefits are relatively easy to estimate. The environmental costs are is expected value of damages due to a spill: E(D)=pD, where p is the probability and D is damages. A back of the envelope estimate is out there, I think. To estimate the probability you could tally the miles of oil pipeline in the U.S. over the years as the denominator and the numerator would be the number of oil spills (=number/miles). This probability could be scaled down by the miles of the Keystone XL pipeline and we'd have an estimate of p. For damages, estimate the average volume of oil pipeline spills in the U.S. and multiply by an estimate of cost of these spills (average fines per gallon over the years?).