Politicians in a democracy face truly crummy incentives: No matter how good or bad their performance, they receive the same salary. What would we expect from a CEO who faced the same reward structure?
As soon as you start talking about paying politicians for performance, however, people start asking: "How would you measure 'performance?'" GDP? GDP+? Average national happiness? Peace? Number of enemies defeated?? Global temperature???
Well, I've got a simple solution. Personally, I don't like it; but I don't see why most people wouldn't. The solution is... [drum roll]:
Pay politicians for approval ratings.
This allows the public itself to decide how much to count anything it pleases. The only tough question is how long to keep paying. If you think politicians frequently sacrifice long-run benefits for short-run gains, you can take the profit out of myopia by continuing to pay after the politician leaves office. If a president's influence fades out after five years, keep conducting approval polls for five years, and keep sending the ex-president his paychecks.
With rational voters, this seems like an elegant solution to many of the political principal-agent problems explored in books like Tim Besley's Principled Agents? My problem with my proposal, as you might guess, is that I think voters are so irrational that the correlation between actual performance and popular approval might well be negative. But obviously I'm in a minority here.
So tell me: If you trust the people, why not pay politicians for approval? If you think that politicians respond to incentives, why not pay billions for an extra percentage point of approval? In fact, even if politicians don't change their behavior in response to incentives, there's still a big benefit to big bonuses: Potential candidates who reasonably expect high approval will be more inclined to run and try harder to win.
Final challenge: Is there any reason to prefer the incentive of re-election to the incentive of payments for approval ratings?