I develop a theory of fundraising competition between candidates whose ability to raise money depends on the policy they will enact if elected. Unlike in existing models of money in politics, there is no exchange of policy favors for donations; donors merely have better information than the electorate about candidates' policy intentions. Even in the absence of quid pro quo exchanges between candidates and donors, I find that fundraising competition may skew policy outcomes away from the center. High spending may signal policy extremity to the median voter, but candidates sympathetic to special interests either spend enough to overcome the bad signal or conceal their intentions by not raising money at all. Consequently, when extremists can raise money most easily, there is a tradeoff between information disclosure and policy representativeness: the chance of electing a centrist is highest when candidates' spending decisions do not reveal their policy intentions. A key policy implication of the model is that campaign finance reforms that raise the opportunity cost of fundraising, such as individual contribution caps, never reduce (and sometimes increase) the chance of electing a centrist, even when the fundraising advantage is associated with centrism rather than extremism.