Q: (me) What are the basic conclusions of the literature regarding overall spending in US elections?
A: (Therriault): With regard to overall spending, Jacobson (1978) was the first to show an effect on vote outcomes, but this effect was mainly present for challengers [in Congressional elections]. In subsequent years, the effect of challenger spending was confirmed, but others also found effects for incumbent spending as well (e.g. Green & Krasno 1988, Erikson & Palfrey 1995, Gerber 1998). The basic takeaway is that spending more is clearly effective for challengers, and probably also matters for incumbents too, but solving the causal direction problems involved makes it very difficult to be really certain of any of these findings.
One problem is we know that winning candidates generally have more money, but whether money helps candidates or is just a signal of unobserved candidate quality [i.e., people give more money to better candidates] is unclear. Another problem is that not only are donors attracted to high-quality candidates just as voters are, but they are also attracted to winning candidates—that is, if money is given in order to get access to elected officials, donors are more likely to give to candidates who are expected to do well, because the expected return is greater. In both cases, we could observe an empirical relationship between winning and having more money for your campaign, without the money actually “causing” the victory.