About 3 years ago, I posted a blog post arguing that the poor don't work due to economic rationality. This was just me playing around with the data and trying to see what I could come up with.
My conclusion back then: by graphing consumption vs income, it turns out that whether you earn $0 or $20k, your consumption is more or less the same. If utility is a monotonic function of consumption, then this means that people who's labor income will lie between $0 and $20k will have no incentive to work. The data I had is hardly perfect - it's just broad aggregates. I don't know where to get a data set which excludes students, the disabled, etc.
Anyway, I forgot about that blog post. It went semi-viral today, so I decided to revisit the topic. Turns out that about a year after I posted it, the CBO issued a report more or less agreeing with it. They compare disposable income to income (rather than consumption to income), but the idea is similar.
Turns out that my rough guesstimates were in the right ballpark. Here is John Cochrane blogging about it. More shockingly, the graphs he shows include cliffs - there are places where earning $1 more can cause you to lose hundreds of dollars in disposable income.