The recode file includes the probabilities for each case in the variables hfs_mod and hfs_sev. These variables have 8 implied decimal places, so you need to divide by 10^8, but that would provide proportions. To produce percentages as in the report, you need to multiply by 100, so in practice the values are divided by 1000000. You can use these probabilities in means to calculate the percentage of de facto households members with a moderate or severe risk of food insecurity as follows:
use "NPPR82FL.DTA"
* 8 implied decimals for hfs_mod and hfs_sev, but less 2 decimals to make it a percentage rather than a proportion - 6 zeros needed below
gen mod = hfs_mod/1000000
gen sev = hfs_sev/1000000
gen wgt = hv005/1000000 // 6 implied decimals for weight
* means
mean mod [iw=wgt] if hv102 == 1
mean sev [iw=wgt] if hv102 == 1
This gives the following results:
. mean mod [iw=wgt] if hv102 == 1
Mean estimation Number of obs = 54,144
--------------------------------------------------------------
| Mean Std. err. [95% conf. interval]
-------------+------------------------------------------------
mod | 12.48814 .1177041 12.25744 12.71884
--------------------------------------------------------------
. mean sev [iw=wgt] if hv102 == 1
Mean estimation Number of obs = 54,144
--------------------------------------------------------------
| Mean Std. err. [95% conf. interval]
-------------+------------------------------------------------
sev | 1.332236 .0349331 1.263767 1.400705
--------------------------------------------------------------