"Partisan Politics and the U.S. Income Distribution (PDF) shows that, since World War II, Democratic presidents have consistently produced greater GDP growth, lower unemployment, and lower inequality than Republican presidents (without producing substantially higer inflation). The differences are remarkable and stand up to virtually every challenge that can be thrown at them (outliers, technology, etc.).
So why have Republicans done so well in presidential elections since 1968? Bartels suggests that the answer is a remarkable difference in economic performance during election years. For reasons that are not entirely clear, Democrats have only managed to deliver 1% growth in pre-tax income on average, while Republicans have delivered about 3%.
A second paper (PDF) expands on this point, showing that voters tend to make their choices based on economic performance in the months before re-election, not on performance over the president's term as a whole.
Finally, his well-known paper "Homer Gets a Tax Cut" (PDF) presents a similarly grim finding: voters largely fail to connect their concerns about growing inequality to the issue of estate tax repeal, and even those who do make the connection still tend to support getting rid of it.
These are depressing but important findings that deserve serious attention from every student of democracy.