Interesting discussion at Freakonomics:

http://www.freakonomics.com/2012/01/12/does-money-really-buy-elections-a-new-marketplace-podcast-full-transcript/

Steve LEVITT: When a candidate doubled their spending, holding everything else constant, they only got an extra one percent of the popular vote. It’s the same if you cut your spending in half, you only lose one percent of the popular vote. So we’re talking about really, really large swings in campaign spending with almost trivial changes in the vote.

RYSSDAL: All right, here’s the thing: Steve Levitt, very nice guy, knowledgeable economist…sadly though, I don’t believe him. Cause if you look, it’s always the guy with the most money who wins.

DUBNER: You’re right; it is almost always the guy with the most money who wins. That is what we know as correlation without cause. So let me explain: When it’s raining out, everybody’s got an umbrella, we know that. Those things are correlated. But you know what, the umbrellas don’t cause the rain, we know that too. Here’s the thing: Winning an election and raising money do go together, but it doesn’t seem as though money actually causes the winning either. It’s just that the kind of candidate who’s attractive to voters also ends up, along the way, attracting a lot of money and the losing candidate, nobody wants to give money to that guy.

[...]

GIULIANI: I tell candidates, it’s always better to be the candidate with the most money, but you can win without it.

RYSSDAL: So to finish the sentence, it’s better to be the candidate with the most money because that means you’re the most popular, right? People like you the best.

DUBNER: That’s exactly right. It’s like saying it’s better to be the radio show host with the most money, but really, what you want to be is the most popular.