Larry Bartels has a paper (pdf) connecting economic growth to political support. Ezra Klein summarizes:
To try to separate whether abnormally good policies led to abnormally good results, Bartels constructed a model that tried to predict election outcomes based on relative growth rates — that is to say, the country’s growth rate subtracted from the average growth rate of developed nations. … there is no evidence that voters are at all interested in whether their economies are performing better relative to other economies in other countries. The only thing that interests them is the absolute performance of their economy. And as you can see in this graph, that interests them rather a lot
The Dish 