Job Creation by Company Size Class: Concentration and Persistence of Job Gains and Losses in Canadian Companies

The statistical observation that small firms have created the majority of new jobs during the 1980s has had a tremendous influence on public policy. Governments have looked to the small firm sector for employment growth, and have promoted policies to augment this expansion. However, recent research in the US suggests that net job creation in the small firm sector may have been overestimated, relative to that in large firms. The first part of this paper addresses various measurement issues raised in the recent research, and uses a very unique Canadian longitudinal data set that encompasses all companies in the Canadian economy to reassess the issue of job creation by firm size. We conclude that over the 1978-92 period, for both the entire Canadian economy and the manufacturing sector, the growth rate of net and gross employment decreases monotonically as the size of firm increases, no matter which method of sizing firms is used. Measurement does matter, however, as the magnitude of the difference in the growth rates of small and large firms is very sensitive to the measurement approaches used. Part one of the paper also produces results for various industrial sectors, and examines employment growth in existing small and large firms (i.e., excluding births). It is found that employment growth in the population of existing small and large firms is very similar. Finally attempts are made to introduce a job quality aspect to the numbers by using payroll distributions rather than employment. The net and gross rates of increase and decrease in payrolls by firm size are found to be only marginally different than those of employment. The second part of the paper looks at concentration of employment creation and destruction within size classes. This is relevant because if growth is highly concentrated, knowing that a firm is small will provide little information about its prospects for growth. Most small firms would grow relatively little, or decline, while a few expanded a lot. It is found that both job creation and destruction is highly concentrated among relatively few firms in all size groups, but it is greater among small and mid-sized companies than large. Finally attempts are made to correlate the performance of businesses over two three-year periods. It is found that knowing that a firm is a high performer (in terms of jobs created) over one period is of only limited value in determining growth in the second period. This is particularly true among small firms. These results suggest that firms which expand rapidly during one period are replaced to some considerable degree by others in the subsequent period.