Same data, different conclusion

Chris Dillow recently commented (emphasis added:

There is a case for paying politicians well, according to this new paper on the effects of variations in salaries paid to local politicians in Brazil:

We find that among municipalities that offer higher wages more educated candidates run for office...Higher wages also encourage politicians to stay in office longer resulting in a higher share of experienced legislators...
Wages also affect politicians' performance. Among municipalities that offer higher wages, politicians submit and approve more legislative bills. This is consistent with an increase in effort induced by the higher future value of holding office....Legislators also signal effort by providing petitions for public works and improvement in public services for their voters. We find that in municipalities with higher salaries, there is an increase in the number of schools, local clinics and an improvement in their infrastructure.

I'd have said that makes a case for paying politicians as little as possible. As an opponent of "managerialism", that's also the conclusion I'd have expected Chris to reach. He does point out it's a possible conclusion, but to me it's an open and shut case.