Counts MSA2013== 0000

I’m trying to understand what can explain the difference between the counts of the variable MSA2013 I get when downloading the microdata and the counts in the case-count view of the documentation. (https://usa.ipums.org/usa-action/variables/MET2013#codes_section)
tab year if met2013==00000 gives me the following counts

census year | Freq. Percent Cum.
------------±----------------------------------
2005 | 761,329 6.67 6.67
2006 | 788,563 6.91 13.59
2007 | 789,812 6.92 20.51
2008 | 787,623 6.90 27.41
2009 | 798,399 7.00 34.41
2010 | 810,360 7.10 41.51
2011 | 868,594 7.61 49.13
2012 | 838,002 7.35 56.47
2013 | 828,969 7.27 63.74
2014 | 828,452 7.26 71.00
2015 | 827,152 7.25 78.25
2016 | 818,143 7.17 85.42
2017 | 824,166 7.22 92.65
2018 | 838,571 7.35 100.00
------------±----------------------------------
Total | 11,408,135 100.00
the documentation says that there are about half of not identifiable areas
Code 0000
Label Not in identifiable area
2018 443,335
2017 436,402
2016 431,828
2015 433,305
2014 431,772
2013 428,338
2012 433,747
2011 449,404
2010 403,959
2009 397,320
2008 393,150
2007 390,693
2006 387,430
2005 356,073
Has anyone run into this issue? What explains this difference?
Thanks
Nathalie

MET2013 is a household-level variable; the frequencies shown on the codes tab are for the number of households in each metropolitan area. If you download an extract of a rectangular file, MET2013 will be tabulated at the person-level. You can confirm this by requesting a hierarchical or a households-only extract. You can then tabulate MET2013 if rectype==“H” in order to filter the observations to only tabulate household records. Alternatively, if you need a person-level tabulation of MET2013, the rectangular file will fit your needs.