I often used the Paul Heyne (now with Boettke and Prychitko) textbook for its many appealing features, one of which is end-of-chapter questions that are far superior to any I have seen in an introductory text. For their discussion of sunk costs, they ask whether casualties suffered in a war can be used to justify staying in the war. The point is that cultural norms may be up against economic thinking. The latter argues for comparing expected (future) costs with expected (future) benefits. But the culture may argue for honoring the sacrifices of the fallen by staying in the fight.
In an "overrated/underrated" discussion of sunk costs, Tyler Cowen cites the culture as a sunk cost not to be underrated. My favorite class discussion head scratcher involves couples tempted to separate but citing all they years they had "invested" in the relationship as getting in the way of a "rational" assessment of whether to stay together. What about all the years already sunk into an academic major connected to an unattractive career (life)? Is it somehow too late to switch?
Private companies shut down the losers. Even public-private ventures throw in the towel on unpromising projects. Airbus may finally cut back on production plans for the A380. Walking away from California high-speed rail will be much more difficult: once a few miles of track are laid, the sunk cost story will be the story. There are no stockholders; bondholders only exist in the promises of the High-Speed Rail Authority which is properly exposed here. (h/t Newmark's Door).
MC=MR static equilibria are standard classroom fare (wonder why do students fail to respond?). But the more plausible model involves dynamics that involve learning from mistakes and error correction. That model includes walking away from mistakes. How else could there be any progress?