Andrew Gelman says that not only is gerrymandering not the cause of partisan polarization, it doesn’t even really make seats safer:
I can’t disagree with Cohen’s first sentence above, but I part company with him after that. When Gary and I looked at the data, we found that redistricting (“gerrymandering”) was not associated with a decline in competitiveness of elections in Congress or state legislatures. Legislative elections have been gradually becoming less competitive, but they are typically more competitive after redistricting.
I’m glad to learn of this empirical result, because I never really understood the theoretical basis of the gerrymander/uncompetitiveness link. Any constituency, no matter how you draw it, is going to have a median voter to whom one can appeal. The shape of the district ought to alter what kind of candidates are viable, but never make it impossible to field viable candidates. I would say that the biggest impediment to competitive elections is fundraising issues. If you had a public financing system that guaranteed a fully funded campaign to the major party nominees in every district, a lot of “safe” seats would suddenly start looking less safe, since it would make sense for both parties to do their best to find candidates well-suited to every district. And that, of course, is why we’re unlikely to see public financing of congressional campaigns.