The pattern of individual, decentralized human evil I’ve discussed in recent posts is an underappreciated argument against the welfare state.  If a substantial fraction of the people around us are money-burners and perverse punishers, how much sympathy will voters have for such folks when they are down on their luck? 
Think of this as a ceteris paribus exercise: If the average poor person were more moral, more kind,  would that raise or lower the political support for the welfare state?  The answer is all around us: Political supporters of the welfare state overwhelmingly focus on sympathetic cases–women with small children, people losing jobs due to offshoring, victims of chronic disease–so they must think that political support for transfer programs depends on a belief that people are basically good.  
But a lot of people are actually just awful.  One example among billions: In a series of studies of male college students in the 1980’s, Malamuth found that about 35% of these students in the U.S. and Canada said they’d consider committing a rape if they knew they wouldn’t get caught; 20% would seriously consider it.  The Malamuth studies sparked the field of “rape proclivity,” a literature that has continued to generate disturbing findings ever since.   And these studies are just detecting those students who are willing to state their proclivities in a survey; the true number is surely higher.
We are, all of us, constantly surrounded by such people.  
Bryan notes that by some moral standards, we don’t owe much to strangers in general; if that’s right, how much less do we owe to strangers who, if they are college-educated males, might have a 35% probability of contemplating rape? And this is just one criminal proclivity, based merely on surveys.  
I suspect that if people were more aware of the awfulness of their neighbors, support for the welfare state would decline.