Confirm how Zewde counted adults in the SCF

[MaxGhenis]

Zewde's appendix states:

The most accessible SCF data are its summary extracts, which we use here.

But also:

Finally, the SCF data also accounts for households with three or more adults, while the tax system only allows for single and two-adult filings. To estimate the economic effects of our program for households in the SCF data with three or more adults, we adopt a simplifying convention: households with three or more adults receive up to $30,000 total for those households with income up to $30,000, declining at a rate of 50 percent above $30,000 to fully phase out at $80,000 in total household income across all adults. Because this category only represents a small proportion of total households, the choice of how to estimate the grant amounts for these households will not significantly affect the overall program cost. Under an alternative assumption that every non-married adult (with two or more adults in the household) file separately, that divides household income in a way that maximizes the refund value, our guaranteed income program cost still remains under $1 trillion.

The summary file only includes marital status and number of children, so the number of adults beyond two is not obtainable. The full file also lacks clarity around whether other fields indicate that the SCF family includes other adults; see PSLmodels/scf#5.