To build on this point, I think it's more likely that builders under-build, and they under-build especially for starter homes for some of the reasons you discuss, but other reasons also.
To take your numbers, the universe of potential buyers for the $160K house is likely greater than for the $225K. But the universe of actual buyers might be small, because a lot of people in the market for the $160K home are likely to have poor credit and thus could have trouble getting financing. I would think that sales of starter homes is more sensitive to economic conditions than middle homes, because the types of folks who are buying the middle homes have more stable employment. Doctors do fine in a recession. Manufacturing companies are likely to lay off line workers, and less willing to cut their R&D staff (though wages might fall a little bit). So building the $160K house is riskier than the $225K house. I don't know if the profit per house would change all that much (econ 10 says the profit should be the same), but the risk profile should.
That said, a lot of this discussion is really a set of empirical questions, and hand-waving, intuition and reasoning from first principles are not terribly good ways to answer empirical questions.
