Could it be that the very factor you are considering, job growth, graduates firms out of the category you are considering at about the pace needed to keep the share constant? (Is the next category up getting bigger?) It makes sense (I think) for the share of the smallest firms to fluctuate in a range, rather than to grow. All that is required is that the number of such firms grow at about the same pace as the labor force. New firms will boost the number of workers in the category, but new hires at existing firms won't, in aggregate, because that should settle out to a pretty constant flow. Yes, the 15th person hired will boost the share of the category, but if the same firm hires 6 more people, all 21 of them are moved up to the next category. Little chunks in, big chucks out, with new firms the only net change. Again, if new firms enter at about the same pace as overall employment growth, the share of workers in firms with 20 workers or less would stay pretty constant. The next larger category would only get credit for the 21st employee, with the first 20 still credited to the lowest category. Lots of employment originating there, but no growth in share. Yes?

I thought this might be the case as well... however the share of employment has decreased for the 20-99 category as well as the <500 category. There might be some jump from the less than 20 to the 500+ group.

I suppose there could also be enough movment up the scale across the board - any way to test this hypothesis?

Percent of Employement...

       <20 20-99 100-499 <500 500+
1999 18% 18% 14% 50% 50%
1998 19% 18% 14% 51% 49%
1997 19% 18% 15% 52% 48%
1996 19% 18% 14% 52% 48%
1995 20% 18% 15% 52% 48%
1994 20% 18% 15% 53% 47%
1993 20% 18% 15% 53% 47%
1992 20% 18% 14% 53% 47%
1991 20% 19% 14% 53% 47%
1990 20% 19% 14% 54% 46%
1989 20% 19% 15% 54% 46%
1988 21% 19% 15% 55% 45%

The stock of employment matches is a funny number, kind of like corporate profits. You take large gross revenues and large gross expenses, and you arrive at a small net profit.

Similarly, with job matches, you get large numbers of new matches, and large numbers of matches that are broken (quits, layoffs, etc.), and you net that out for total employment.

So, what I think is going on here is that small firms *create* more jobs, but they also *destroy* more jobs. So the stock of jobs in small companies does not rise nearly as fast as the job creation figures would suggest.

At least, that's my hypothesis.

You're, of course, right that there is both job destruction as well as creation.

However, the data claim to be "net" changes - it subtracts losses from "firm deaths" and gives a "net change" number. For example, the net change is negative in 1990-1991.

So, I'm not sure if your hypothesis is correct unless there is something else hidden in the data.

The figures show that the "net change" in new firms is very small compared to the gross births and deaths. This makes the data very susceptible to error. For example, I would conjecture that the response rate to surveys is much lower among dying firms than growing firms. A small response bias that favors growing firms would result in a huge error in the overall data. I'm sure that the statisticians try to correct for this, but it must be quite difficult to do. I just don't think we have the instruments to precisely estimate the net job creation of small business taken as a whole.

As a small business employer (25 years with 18-28 employees), I confess to having little familiarity with terms like "firm births" and "firm deaths," but I would like to offer the following considerations based on my experience:

1. Do the data go beyond job creation/destruction at the time of a firm's birth or death to capture trends in job downsizing and job growth which occur without a "birth" or a "death"? Changes in the size of a small company are likely to be much more frequent because they will be more responsive to seasonality and to local economic trends than are those in the larger firms.

2. There seems to be no distinction in these data among job classifications. Smaller firms are more likely to create data problems in categorizing their employees as part-time, full-time, hourly, salaried, etc. because there will be little consistency in each responder's definition of these terms. The surveys which were presnted to me for completion never defined these terms even though they asked for the data.

My expectation is that smaller firms will create more jobs, but that these jobs will generally be shorter-lived than those in larger firms and they will be accompanied by fewer direct monetary benefits. However, the jobs in smaller firms will be accompanied by more indirect benefits such as discounts on the firm's products, free use of company assets, etc. We end up with an almost impossible attempt at sorting the apples from the oranges.

I just happened on to this site searching for information on small business and job creation. Since this seems to be your topic I would like to ask a probably dumb question----why has the SBA(Small business administration) been allowed to wither on the vine? What would happen if just a small portion of the huge "tax cut" proposed was granted to the SBA to help create jobs. Correct me if my thinking is flawed please but would I be correct in saying this:
1. Every dollar that is provided to an entrepeneur to start a new business is pumped directly into the economy.
2. I realize there is a very substantial failure rate but those that survive may become the next Microsoft, or Sun Microsystems etc.
3. Most of the existing companies are still downsizing----not expanding

How does this create jobs?

Just wondering how mergers and acquisitions are accounted for in the numbers? The 90s were a boom in this area as firms like Cisco, CMGI, and so on scooped up small start ups. In other words, would employment gains by merger skew small firm percentages down and large firm up?

I'm not sure I undestand your data. The net job creation you refer to looks to me like net firm creation. The number of jobs for firms with less than 20 employees in 1999 was 20.4 million on your spreadsheet, yet lower down, you use the number of firms to calculate the increase of 53%. Am I missing something here, or are you saying that firms that account for 18% of employment account for 53% of the change in the number of firms? As I read the data, they only accounted for 4% of total employment growth in 1999.