At Econlog, the very sharp Garett Jones makes an argument for paying politicians more:
There’s some evidence that when it comes to politician quality, you get what you pay for; Besley finds that higher pay for U.S. governors predicts governors with more experience in politics, and Ferraz and Finan look at Brazilian data and find a slower revolving door and better educated politicians in regions where politicians get better pay. But alas the egalitarian ethos in democracies makes it difficult to raise the pay of politicians.
But there’s a countervailing effect of high salaries for politicians: they increase careerism. With high salaries for politicians, you’re more likely to get candidates who give the voters what they want so that they can get (re-)elected. And one of the themes of Jones’s post is that the voters are ignorant and excessively egalitarian: we shouldn’t always give them what they want. We need politicians who are intelligent, informed, and public-spirited. High salaries get us more of the first two and less of the last.
What else does the evidence suggest? In the American states, governments that pay legislators more and generally have more professionalized legislatures have higher government spending. Neil Malhotra has found good evidence that the causal arrow goes from spending to professionalism rather than the other way around. However, his study, for all its sophistication, has some evidentiary holes, and I believe the last word has not been spoken. From my own observations of the highly deprofessionalized, low-paying ($100 a year) New Hampshire legislature, I would say that it attracts candidates who are ideologically motivated but not careerist. They deviate significantly from the views of the median voter, for good or ill.