Comparing Household Survey with Payroll Numbers

A quick addition to the last post. One of the concerns was the large difference between the payroll numbers (114,000) and household survey numbers (873,000) (for a detailed definition check here). I am simply going to point out that the difference is not that big comparing to previous periods. Here is a graph showing the difference in the month-to-month change in the employment data where I have taken the difference between the household survey data and the payroll data.

As expected, the graph still shows a lot of noise and in absolute terms 2012 does not appear to be an anomaly. 

Month	Difference in Housing Survey and Payroll
2000-01-01	1788.00000
1960-04-01	932.00000
1948-04-01	931.00000
1990-01-01	913.00000
2003-01-01	896.00000
2002-02-01	883.00000
2001-09-01	848.00000
2012-09-01	759.00000
1954-02-01	725.00000
1983-08-01	708.00000
1991-04-01	669.00000
1953-01-01	665.00000
1960-11-01	659.00000
1948-06-01	652.00000
2002-09-01	652.00000
1970-10-01	617.00000
1983-06-01	613.00000
1956-07-01	609.00000