search

Treasury-Analytics-and-Insights / emtr

Nascent Python version of IncomeExplorer
1 stars 0 forks source link

scaling isn't quite right on income composition plot - suspect weeks_per_year issue #23

Closed corypdavis closed 1 year ago

corypdavis commented 1 year ago

default family: WEP: 696, should be 700 FTC: 18674, should be 7121 + 5802 + 5802 = 18725 BestStart = 3622, should be 3632.

corypdavis commented 1 year ago

OK, it turns out that this issue was apparent in the R version. Basically values look inconsistent with legislated parameters (in particular annual ones like WfF) because they are all scalled down to weekly payments in different ways. The income composition plots just scales these up again by a weeks_in_year factor (which was 52). Propose instead of 52, which systematically undercounts, we use the existing wks_in_year function, which accounts for leap years.