On my last post I got some great comments! Jeff "Pantagraph Trolleypole" Wood pointed us to Pushkarev and Zupan suggesting that commercial density is more important than residential density back in 1977; Jeff summarized that argument in 2010. So that's where a lot of this comes from.
Richard Layman points out that some parts of the transit network can reinforce others, and that it may therefore be valuable for apparently unjustified transit routes and runs to be cross-subsidized by "popular" routes, or even by the government. Jeff also pointed to the value of high-speed rail in inducing dense development near stations, and I think if we put these together we get something that Germà Bel's analysis misses: that an "unprofitable" high-speed rail line can be worth subsidizing if it gets people to downtown stations where they will walk and ride transit instead of driving.
Bel himself left a comment pointing out that any transportation investment yields private benefits (which should probably be paid for by the user) and public benefits (which could be paid for by the government), and pointing us to an interesting study of the new Italian high-speed rail network (PDF). I think that's very important, but I would add that the public and private benefits are not a matter of universal agreement, and especially with the public benefits there will be people who disagree about the relative value of transit cross-subsidy or economic development or emissions reductions. The debate is not just a matter of how much a project affects these outcomes, but how much the outcomes matter.
With that in mind, let me try to clear up a few more things about return on investment. If our goals are to induce economic development then we'll be looking at measures like levels of employment and tax revenue generated by that development. If we don't care about any of that and we just want to make sure that this project doesn't bankrupt the state, we'll be focused on capital and operating outlays. Either way, density of development plays a role. Just as importantly, though, we need to look at the transportation system as a whole consisting of redundant bus, rail and private auto networks, and figure out the most cost-effective way to make use of it. It's idiotic to declaim the waste on high-speed rail while ignoring the ROI of the multitude of inefficient highway expansion and rehabilitation projects. As Chuck shows, development density improves the ROI of both road and transit infrastructure.
As you can see at the top of the page, my goals are to increase access and improve society while reducing pollution and carnage and avoiding resource depletion. Because of that I may look at individual measures like access to jobs and services, pollution, fatality and injury levels, and rate of depletion of the various natural resources used by transportation and development. Many of these metrics are sensitive to the density of residential, commercial and industrial development.
However, other than access, all those goals require getting people out of their cars. This means that I can look at VMT reduction as a long-term goal, and increasing the mode share of transit and walking in the short term. More importantly, there's a cycle of government and private investment in transit. The more transit ridership there is, the less subsidy will be required, and the more money that will be available for expanding the transit network. In addition, the more transit ridership there is, the more political support there will be for government investment in expanding the transit network. Conversely, the less driving there is, the less political support there will be for driving subsidies. Transit mode share is really key here, more important than any of the individual measures.
Transit mode share, in fact, is where density is least relevant. This may seem surprising, but only if you believe that density is the only way to control the relative value that people get from various modes. I'll talk more about that soon.