Read on Twitter
If stimulus checks were larger, would people spend more?
Natural starting point: how were the checks that went out in April and May spent?
tldr: we have no idea
estimates all over the place. afaik no reconciliation.
(wonky thread on 3 papers w conflicting results)
Natural starting point: how were the checks that went out in April and May spent?
tldr: we have no idea
estimates all over the place. afaik no reconciliation.
(wonky thread on 3 papers w conflicting results)
1) Census Pulse Survey asked "did or will you use it [Economic Impact Payment] mostly to pay for expenses, mostly to pay off debt, or mostly to add to savings?"
Expenses [Spending]: 70%
Debt: 16%
Savings: 14%
https://www.census.gov/data/tables/2020/demo/hhp/hhp7.html
![1) Census Pulse Survey asked "did or will you use it [Economic Impact Payment] mostly to pay for expenses, mostly to pay off debt, or mostly to add to savings?"Expenses [Spending]: 70%Debt: 16%Savings: 14% https://www.census.gov/data/tables/2020/demo/hhp/hhp7.html](https://pbs.twimg.com/media/EqX2pWmXYAACcvw.jpg "1) Census Pulse Survey asked \"did or will you use it [Economic Impact Payment] mostly to pay for expenses, mostly to pay off debt, or mostly to add to savings?\"Expenses [Spending]: 70%Debt: 16%Savings: 14% https://www.census.gov/data/tables/2020/demo/hhp/hhp7.html")
Expenses [Spending]: 70%
Debt: 16%
Savings: 14%
https://www.census.gov/data/tables/2020/demo/hhp/hhp7.html
Only analysis I know of these survey questions is by @GarneratBLS @adamsafir @JakeJSchild https://www.bls.gov/opub/btn/volume-9/receipt-and-use-of-stimulus-payments-in-the-time-of-the-covid-19-pandemic.htm
2) Survey of Nielsen respondents by Coibion, Gorodnichenko, and Weber
Spending: 15%
Debt: 52%
Savings: 33%
Table 2: https://faculty.chicagobooth.edu/-/media/faculty/michael-weber/stimulus-spending.pdf
Spending: 15%
Debt: 52%
Savings: 33%
Table 2: https://faculty.chicagobooth.edu/-/media/faculty/michael-weber/stimulus-spending.pdf
3) App data for very low-income households by Baker, Farrokhnia, Meyer, @PagelMichaela, and Yannellis
Spending MPC of 37% in sample. Spending MPC of 27% when reweighting sample on observables to look like the US. https://drive.google.com/file/d/19KaPY-xV27feRZF74EBby3DZ3RW9K3Bp/view
Spending MPC of 37% in sample. Spending MPC of 27% when reweighting sample on observables to look like the US. https://drive.google.com/file/d/19KaPY-xV27feRZF74EBby3DZ3RW9K3Bp/view
in addition, 3) has important shortcomings: estimates only go out 3 weeks, yet we expect spending response to be delayed by shutdowns.
also, timing is not random (lower inc hhs get refunds first) so comparing early and late recipients may not be a valid research design
also, timing is not random (lower inc hhs get refunds first) so comparing early and late recipients may not be a valid research design
So is 15% spent or is 70% spent?
It is completely unclear.
The Census Pulse Survey is almost certainly better on representativeness than Nielsen. But Nielsen survey was able to go more in depth and respondents are more motivated to answer questions about consumption.
It is completely unclear.
The Census Pulse Survey is almost certainly better on representativeness than Nielsen. But Nielsen survey was able to go more in depth and respondents are more motivated to answer questions about consumption.
Further, it would be really helpful for papers (1), (2), and (3) to be in explicit dialogue with each other.
Compare estimates, try to think about why they might be different from one another. But the papers have either no mention of the other estimates or a cursory citation.
Compare estimates, try to think about why they might be different from one another. But the papers have either no mention of the other estimates or a cursory citation.
Caveats & notes: All three sets of estimates are for people getting an Economic Impact Payment. I wish more of the authors were on twitter. Please tag any author in the comments who I should have tagged but missed.
4) I missed this paper by @EzraKarger @Aastha_rajan. It focuses on a very low-income sample of EIP recipients using prepaid card data. https://www.chicagofed.org/publications/working-papers/2020/2020-15
This paper has a much more detailed discussion of the other EIP estimates out there, which is great. However, the paper does not acknowledge or attempt to reconcile the differences in the estimates.
