The "Bread and Peace" model may yield decent predictions of election results, but we still must consider aspects of politics that are not well captured in this particular model. The selection pressure you describe is real enough, but it is hardly the only thing that affects policy and politics, as the links I provided describe.