Extremely strange Household Employment Survey Data

The Household 
Employment Survey,
reported yesterday,
which is more accurate 
in the months before 
a recession starts, 
than the Payroll Survey,
which gets the headlines,
reflects 103,000 
fewer jobs 
in April 2019,
than in March 2019,


... and 300,000 
fewer jobs
in April 2019, 
than in 
December 2018.


The Household Survey
is about 50% more volatile
than the Payroll Survey,
so one month (April)
may mean little,
but it never needs 
to be revised,
because 
it's based 
on phone calls 
to real people.

The Household Survey 
includes self employed 
(The Payroll Survey does not), 


The Household Survey 
does not double-count jobs
when a person has 
two or three part-time jobs.
(The Payroll Survey does).

The Household Survey 
is more accurate
in the months just before 
a recession begins,
but all employment surveys 
are LAGGING indicators.

The Payroll Survey 
tends to need 
huge negative revisions
in the months
just before, 
and just after, 
a recession begins.