The economy has slowed somewhat since mid-2006. Tight labor markets, which could result in upward pressure on wages and employment costs, remain one of the key concerns for the Fed. There are two key employment surveys produced by the U.S. Bureau of Labor Statistics (BLS). What are they?  What are some key differences in the surveys and what are some different insights into whether the labor markets are loosening any?

A comparison of the two surveys

The U.S. Bureau of Labor Statistics releases employment data generally on the first Friday of each month. The “employment situation” actually has employment data from two surveys — the household survey and the establishment survey. The household survey is a door-to-door survey based on about 60,000 households. This survey is more formally called the Current Population Statistics survey which is conducted by the Census Bureau but analyzed by the BLS. The survey covers the farm and non-farm sectors and includes self-employed and some unpaid workers. Civilian and non-civilian workers are included but only for those age 16 years and older. For the number of employed, each worker is only counted once. The survey includes information on unemployment status. The unemployment rate is the highlight of this survey.

The establishment survey is also called the nonfarm payroll survey — more formally the Current Employment Statistics survey. It is a survey of about 400,000 businesses.  These businesses report their payroll numbers, earnings, hours worked, among data to the Labor Department. The highlight of this report is the change in payroll jobs each month. The survey covers the civilian nonfarm sector only. There is no age limitation.  If an employee works for more than one establishment, each job is counted in the job tally. The self-employed are not included which is the biggest difference between the two surveys. The difference related to agricultural workers is relatively minor. Workers on strike are counted as not being on payroll whereas the household survey counts strikers as employed.

Essentially, the household survey can give us employment comparisons largely based on socio-demographic factors while the establishment survey provides industry oriented data.

Recent broad trends

How do the employment numbers compare recently?  During the current expansion household employment generally has risen faster than payroll employment — at least through the end of 2006. 

Coming out of the last recession, household employment clearly outstripped payroll employment growth during 2002 and 2003 but both were essentially equal in 2004 and 2005. But household employment accelerated relative to payroll job growth during 2006.

For example, during 2006, the average monthly increase in household employment was 262,000, compared to 189,000 for payroll jobs. According to the household survey, employment growth was very strong and helped to explain a low unemployment rate. Household employment growth slowed sharply in early 2007, even dropping 468,000 in April. Currently, both employment measures are soft.

Could some of the differences between the household survey and payroll survey explain the different growth rates in alternative measures of employment? Examining the shares of key components of household employment can give insight into this possibility. Three differences between the two surveys are that the household survey includes the farm sector and also includes the self-employed and unpaid family members. These concepts overlap, so two separate charts can best illustrate their relative importance.

As seen above, the farm sector accounts for only 1.5 percent of total household employment. Changes in farm employment are not going to have any notable impact on the overall trend in household employ.

However, for 2006, the self-employed made up 7.3 percent of total household employment. Unpaid family members who were considered workers were only 0.1 percent of the total. The key difference between household employment and establishment survey employment is the self-employed component.

Does real growth explain changes in growth of self-employed?

One of the theories behind strength in self-employment growth is that payroll job growth is too weak and laid off workers are forced to become self-employed. However, strength in self-employment growth does not always coincide with weak real GDP growth and weakness in self-employment growth doesn't always coincide with strong growth. Over the 2003-2006 period, overall economic growth was healthy but self-employment growth swung between healthy growth and declines. Changes in growth rates for the self-employed seem to be affected by cycle specific factors. For example, the increased use of outsourcing overseas by companies instead of contracting in the U.S. likely has reduced some opportunities both payroll and self-employment jobs while at the same time forcing laid off workers to consider self-employment.

Excluding the self-employed, household employment growth has still been running a little ahead of establishment employment growth.

Job opening rates and self-employment

Growth rates in self-employed do seem to be affected by “job opening rates” — but the relationship is not consistent  The Labor Department keeps data on job opening rates by industry as a percentage of existing jobs. The below compares the growth rate in overall self-employed to the job opening rate for professional and business service sectors. These are job sectors from which it is believed there is significant self-employment. 

Regarding the impact of job openings on self-employment, there is a lag in the impact on year-on-year growth rates. The high hire rate in 2000 seems to have lowered self-employment growth in 2001, and the low hire rate in 2001-2002 appears to have boosted self-employment growth in 2002-2003. What clearly does not match the pattern is the rise in self-employment in 2006 since the hiring rate had been relatively high since 2005.

Bottom line

To some degree it is easy to see some of the differences in growth rates between the establishment survey and the household survey — at least in terms of components.  However, seeing the differences is not the same thing as understanding the causes for the differences which appears to be the recent slowing in self-employment. Nonetheless, both surveys are pointing to marginal slowing in labor markets.