Household vs payrolls? Which number's closer to the truth?

Just a quick follow up from my questioning of the significant difference between Wisconsin's payroll "jobs" survey, and the household "employed/unemployed" survey.

Let's add a wider perspective of other states into this equation, based on Friday's release of the state-by-state jobs figures by the Bureau of Labor Statistics. If you go by the payroll figures over the last 12 months, Wisconsin is in its typical place during the Age of Fitzwalkerstan- in the bottom half of job growth in the Midwest, and well behind our neighbors in Minnesota.

Change in private sector jobs, May 2016 - May 2017
Minn +1.86%
Mich +1.64%
Ind. +1.36%
Iowa +1.31%
Wis. +1.22%
Ohio +1.09%
Ill. +0.70%

But if you look at the household survey, the numbers are very different. This chart shows the 12-month difference in jobs reported in the payroll survey and number of people described as "employed" in the household survey, and you'll notice that while some states are right in line, other states like Ohio and Wisconsin have major gaps.



Which goes back to my point from a few days ago- why is there such a huge gap in these two surveys in Wisconsin, and which number is likely to be correct? one thing that I want to see is if this disparity is reduced over the next few months.

Seasonal vs. non-seasonal totals, March - May 2017
Payroll jobs
Non-seasonal total +69,300
Seasonally adjusted +11,700

Household employment
Non-seasonal total +54,300
Seasonally adjusted +29,500

Yes, it looks like Wisconsin is adding jobs above and beyond the typical increase in April and May as the weather warms, but can you explain how this is such a difference? When the total number of people reporting as employed becomes more than twice as many people working when the "seasonal adjustment" hits? That has to be changed in the upcoming months, and let's see what happens to those unemployment numbers in the coming months- particularly when we constantly hear businesses complaining about the inability to find workers in a time when workers are usually added.

The answer will likely become clearer in the next 2 1/2 months leading up to the next release of the "gold standard" Quarterly Census of Employment and Wages (QCEW) in early September. But there are jobs reports in the meantime to come out, and we need the picture to clear up soon enough.