The Atlanta Fed's macroblog provides commentary and analysis on economic topics including monetary policy, macroeconomic developments, inflation, labor economics, and financial issues.

Authors for macroblog are Dave Altig, John Robertson, and other Atlanta Fed economists and researchers.

« The GDP revisions: What changed? | Main | Lots of ground to cover: An update »

August 18, 2011

The new firm employment puzzle

Last week, John Bussey of the Wall Street Journal identified some discouraging statistics from the U.S. Bureau of Labor Statistics "Business Employment Dynamics" (BED) program concerning the number of new establishments (a specific physical location like a store) and the number of jobs from new establishments. These data are replicated in the following chart, which shows both a steep decline in the number of establishments over the last few years but also a decade-long trend of a declining number of jobs coming from new establishments.

BED New Establishment Data Annual

Not only has the number of new establishments declined, but the average size of new establishments has also tended to decline over time.

BED New Establishment Data: Average Size Annual

If small businesses, or more specifically new small businesses, are an engine of job growth in the United States, that particular engine has been getting less powerful. Between March 2009 and March 2010, new establishments generated 1 million fewer jobs than over the period from 2005 to 2007. About 85 percent of that decline was related to a reduction in the number of new establishments. The other 15 percent was attributable to the smaller average size of those new establishments. In contrast, the 1 million fewer jobs per year coming from new establishments between the years 2000 and 2005 were all attributable to a decline in the size of the establishments—the number of new establishments per year actually rose slightly over that period.

The BED program also maintains a quarterly series on establishment births (the annual data are for the year ending in March). The quarterly data show that the number of new establishments rebounded over the latter half of 2010 to a pace comparable to the late 1990s. But the associated number of jobs did not increase proportionately.

BED New Establishment Data Quarterly

So while there appears to have been a healthy pick-up in the number of new establishments in late 2010, the gradual march toward ever-smaller new establishments seems to be continuing.

The changing industrial composition of the economy doesn't seem to be the explanation for the declining establishment size trend—the pictures look basically the same if sectors such as manufacturing and construction are excluded. Perhaps it is that new establishments are simply more able than older establishments to adopt new technologies and processes that reduce the demand for labor because they have no legacy employment or capital to deal with.

How robust are these findings? The BED data do have some drawbacks. For example, the BED data are extracted from the administrative unemployment insurance records for businesses that have payrolls (employees). This covers the vast majority of workers in the United States (about 98 percent of employees on nonfarm payrolls and 94 percent of total employment) but not the majority of businesses (U.S. Census Bureau figures for 2006 report that there were about 3.5 times as many firms without employees other than the owner(s) than firms with employees). Also, these data do not distinguish between a new location that is part of a new firm and a new location that is part of a larger multi-establishment firm (like a national chain).

The Census Bureau maintains a related, but different, data set—the Business Dynamics Statistics (BDS)—that allows distinguishing between new firms and new establishments. However, the latest publicly available data only go through the year ending March 2009. The BDS data for new establishments from new firms and the jobs from those new establishments are shown in the next chart. Like the BED data, the BDS new firm data show that the number of new establishments peaked in 2006 and has been trending lower since.

Number of New Establishments

However, the two data sources differ in terms of job creation: the BED new establishment employment data peaked in 1999 and has been declining steadily since, while the BDS new firm employment data peaked in 2006 and was relatively stable prior to that.

Number of Jobs at New Establishments

Compared to the BED data, the average size of new establishments from new firms in the BDS data has changed relatively little over time—the swings in the number of establishments and employment in the BDS data move about in proportion to each other.

Average New Establishment Size

Interestingly, the BDS data on the average size of new establishments at firms of all ages (new firms as well as older firms) appears to be more cyclically sensitive than the new firm data, suggesting that older firms respond more to prevailing economic conditions than new firms do. Neither of the two BDS series displays the secular trend apparent in the BED data over the last decade.

Why do inferences about new establishments from the BED and BDS data differ? I don't know. This 2009 paper by U.S. Bureau of Labor Statistics economists Akbar Sadeghi, James Spletzer and David Talan tries to reconcile the differences but concludes that a definitive answer would require linking the data series together to compare the measures for matched establishments.

There is plenty of scope for investigating the dynamics of new business formation using data at a U.S. Census Bureau Research Data Center (RDC). The newest RDC will be located at the Atlanta Fed, which has partnered with six other research institutions to apply for and be approved to operate the 13th RDC location in the United States. The Atlanta Census Research Data Center will be a secure location where approved researchers will have access to restricted microdata in order to investigate questions like these. The Atlanta RDC is scheduled to open in September. For researchers in the Southeast interested in exploring research opportunities at the Atlanta Census Research Data Center, contact Melissa Banzhaf. For more information about U.S. Census Bureau Research Data Centers, and the Atlanta Census Research Data Center in particular, see here.

Update: Hat tip to Jim Spletzer at the Bureau of Labor Statistics, who pointed me to this paper by E.J. Reedy and Bob Litan, which was published by the Kauffman Foundation in July. Reedy and Litan show similar patterns in the aggregate BLS and BDS data, as I do, and also demonstrate that the shrinking size of new establishments is not made up for in later life. New firms are getting smaller on average and tend to stay smaller over their life. They posit as explanations increased firm-level productivity and shifting occupational needs related to increased use of information technology and increased globalization.

John Robertson By John Robertson, vice president and senior economist in the Atlanta Fed's research department

August 18, 2011 in Employment , Small Business | Permalink


TrackBack URL for this entry:

Listed below are links to blogs that reference The new firm employment puzzle :


You are a senior economist and don't understand the long term trend from skilled labor to automation? Or the capital required in today's business landscape small or large?
Try starting a consumer driven business that requires access to store shelves? or B to B calling on any major corporation that requires extensive ISO documentation along with bidding against better capitalized competitors. Internet driven E-bay style companies are mostly what is happening in small business otherwise the capital requirements far exceeds what most Americans can risk.
I am retired from a career in manufacturing that started in the early 70's and have experienced first hand the country's changing business and labor market.

Posted by: Ron Caldwell | August 19, 2011 at 11:09 AM

I would like to comment on this previous post. I think one must consider all of the so called independent contractors working as freelancers that are paid by the job not by the hour this situation is ripe for abuse. This could be destorting the jobs picture.

Posted by: dennis the menace | September 06, 2011 at 02:55 PM

I think you have hit the nail on the head here. Newer small establishments are able to stay more focused on a particular sector and make decisions quicker without all the redtape of a larger organization.

Posted by: Recruitment Agencies Auckland | December 20, 2011 at 03:05 PM

Post a comment

Comments are moderated and will not appear until the moderator has approved them.

If you have a TypeKey or TypePad account, please Sign in

Google Search

Recent Posts



Powered by TypePad