The main difficulty with CBA is not the framework. Its the lack of information on non-market values such as the value motorists place on their own time - a critical parameter on the benefits side of CBAs for transport infrastructure projects.
I wonder whether some kind of demand-revealing referendum process would be helpful, perhaps not universally but in some cases. This is a clever game devised by Ed Clarke and nicely explained by Chris Dillow in the context of the decision to build a trident submarine.
The basic idea is that everyone gets to vote a $ amount for clearly stated outcomes. The game is arranged so that
- voting your actual valuation is the only sensible thing to do;
- mostly you won't have to pay your vote valuation; and
- you only pay if you are in a small minority that gets its desired outcome by voting more $
Why not give this a go? I think the biggest barrier might be legal: how do you credibly commit to getting the ca$h out of people if necessary? Unless this is a credible threat, the truth-revealing mechanism falls to pieces.