The 99th percentile ("top 1%") of wealth by household seems to have been be $8.4M in 2007, according to https://economix.blogs.nytimes.com/2012/01/17/measuring-the-..., the most recent calculation I could find from a quick Google before heading to work - I'd love to see these calculations done with the raw data at https://www.federalreserve.gov/econres/scfindex.htm . From the summary on that page, they surveyed some 116 million households in 2007.
Suppose you tax all holdings at the top 1% of wealth by 1%/year, which is significant but not enough to wipe you out - after 50 years that leaves you with 60% of your original savings, assuming you weren't investing it.
That yields as a minimum $97T in tax revenue per year ($8.4M * 1.16M * 1%), and almost certainly significantly more because there-s a short tail of people with much more than $8.4M net worth. But if you split even this much among the 5% of households with "very low food security", that's $16,800/year. That's a lot of food.
(An actual scheme would have some sort of progressive tax, also, not a discontinuity at a certain dollar value)
Total US GDP for 2016 was ~$18T.
It seems a bit unrealistic to me to think we can get 5x GDP in tax revenue each year.
So what is the catch?
I'm still interested in doing or seeing an analysis with the full Fed dataset.