No one makes, say, $15k a year and decides "Hey, this is barely sustainable. Let's just coast here for a while."
(and a $300/month benefit is *substantially* less than that!)
Best evidence we have is that unconditional benefits have a tiny effect on labor supply. @mioana's work suggests that on average you see a $1 reduction in market income for every $10 in unconditional benefits.
In other words, you'd that a $3,600/year child benefit would reduce earned income by around $360/year.
That's 24 hours for a $15/hour job.
I'm ok with people taking 3 more days off than they would in the counterfactual.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
Saying "incentives matter" doesn't have to imply anything negative about people living in poverty. In general, the idea that people are very sharp and goal-oriented (which is what I generally understand 'incentives matter' to mean) is good.
That said, I think this language is often sort of crudely appropriated and used in a misleading way.
This is particularly the case with discussions of welfare reform, where people use this kind of language to make the exact opposite claim that "incentives matter" should tell us.
A big problem with #2 is that the aggregate data is messy enough that it is easy to paint any picture you want. Was 1990-era increases in labor participation due to:
1- Continued secular trends 2- Welfare reform 3- A hot business cycle 4- EITC expansion
A lot of what *looks like* disincentives are often caused by program *phase-outs*, not the programs themselves. I've talked to people who have refused promotions that would mean they lose Medicaid eligibility, for example.
The economics 101 approach has a lot of insight here. People make decisions at the margin, and getting a flat child support payment has at worst a small impact on the decision from that perspective. ideas42.org/wp-content/upl…
Not sure if MattY touches on this, but I've been arguing that a lot of the evidence for the proposed greater sustainability of universal programs seems to have somewhat shaky causal reasoning.
ie, here's an example from a paper @jdcmedlock shared a while back.
Note that:
1 - The effect is tiny. Going from the least universal state to the most universal only increases support by 15%!
2 - It seems very likely that *most* of the causation here is culture=>institutions
Better argument for universal programs is probably just that means-tested programs end up being kinda janky.
Very easy to have lots of programs with weird overlapping phase-outs such that folks end up in poverty traps. Means tested programs are harder to design and operate.
1.) There's no reason to have an arbitrary $1 t limit.
2.) To the extent that you want the bill to be smaller (regardless of the validity of such claims) there's still a good case for some sort of universal payment, even if less than $1200.
One thing we know about the implementation of CARES is that it's spotty and time lagged. Lots of people who should have gotten FPUC didn't, or got it months later than they should of.
That's true of checks too! But we have the mechanisms in place to send that money out, and can learn from any problems in the CARES implementation.
The argument is that (Biden 2020 vote - Clinton 2016) vote is not normally distributed, which suggests fraud.
But:
1.) There's no particular reason for it to be normally distributed. Not every phenomena is normally distributed.
2.) Eyeballing it, it *is* normally distributed.
Oh jeez, it seems likely the "weird outliers" are entirely the result of bad data cleaning.
For example, "Pleasant Ridge" was two precincts in 2016 (691 and 670 votes for Clinton, respectively), and one precinct in 2020 (such that "Precinct 1" had 1605 votes).