Establishment Size and Employment Growth in the Canadian Urban System 1991-1994: An Exploratory Analysis
Shearmur, Richard G., J. Coffey, William, Canadian Journal of Regional Science
Richard G. Shearmur and William J. Coffey
Departement de Geographie
Universite de Montreal
Montreal, PQ H3C 3J7
It has become accepted wisdom today that "the role of creating jobs has moved from big companies to small business and the self-employed" (Globe and Mail 1996a). Probably the ultimate source of the accepted wisdom can be traced back to Birch's work in the late 1970's, in which he reported that 66% of net new jobs created in the US economy between 1969 and 1976 were generated by firms with less than 20 employees (Birch 1979). This result did not go unquestioned. A study commissioned by the US Small Business Administration for the 1978-1980 period found that firms of 100 employees or less only created 38% of all jobs: the same period was subsequently analysed by Birch using the same data, who obtained a figure of 70%! (Giaoutzi et al. 1988). Storey and Johnson (1987) examined these conflicting results and reported that the difference stemmed from the different treatment of firms not included in the data base: in their view, 50% of new jobs were created by companies of less than 100 employees over the period covered, while the latter employed only 38% of the labour force.
Whether or not such statistical observations could truly be construed as supporting the contention that small businesses created a disproportionate share of new jobs, "during the 1980's (the observations) had a tremendous impact on public policy" (Picot et al. 1994). The message taken by many "economic development officials (in the USA) is that job-generation begins at home ... (and that) ... efforts are best concentrated in helping local businesses expand and encouraging potential entrepreneurs to establish businesses" (Eisinger 1988). Indeed, with average unemployment rates in Canada and in many European countries rising to a seemingly irreducible 10% or so over the 1980's, small businesses have been seized upon as generators of employment.
One of the consequences of such research has been a fundamental reappraisal of approaches to regional development policy. Whereas throughout the 60's, 70's and early 80's Canadian regional policy concentrated on a variety of top down approaches such as infrastructure provision, capital investment grants, and incentive packages (Savoie 1986), there has been an increasing trend towards the promotion of local development initiatives (Economic Council of Canada 1990c; Fondation de l'Entrepreneurship 1994). (1)
Needless to say, statistical evidence alone is unlikely to have brought about such fundamental policy shifts. The increasing budget problems faced by most western governments have reduced their capacity to finance regional policy on a grander scale, and local development is perceived as a low cost alternative. It also fits well with the current anti-government bias in popular opinion, as local development is seen as a bottom-up approach which empowers individuals and municipalities at the local level. Thus, while local development in general, and specifically the aspect of it which stresses small businesses and entrepreneurship, can be presented as a rational response to a series of empirical observations concerning employment creation, it can also be seen as an opportune policy initiative in keeping with the generalised move towards liberal economic management experienced by western countries in the `80's and early `90's.
In the light of these developments, this paper first reviews the evidence that small companies create more jobs than larger ones, and examines the reasons why this may be so. While the evidence does appear to point this way, the policy implications are unclear for a number of reasons which are set out below. After reviewing this evidence, the results of our analysis of 59 Canadian cities are presented.
This exploratory analysis introduces a geographic dimension to the analysis of Small and Medium Enterprises (SMEs) which has hitherto been largely ignored in the literature. Indeed, while this dimension is stressed by geographers (Openshaw 1981; Bourne 1991 etc...) economists have only recently begun to take into account the spatial dimension of the phenomena that they study (Krugman 1991; Martin and Sunley 1994; Vedder and Gallaway 1996). (2) On the assumption that there is a link between establishment size and employment growth, as posited by much recent research, and that, furthermore, it is the smallest establishments which tend to stimulate employment most, one should observe that geographic areas with higher concentrations of smaller establishments benefit from faster growth rates. In addition, it can be expected that areas which experience faster growth in the relative number of small businesses will also benefit from faster employment growth. It is these hypotheses which the following empirical analysis seeks to verify within the Canadian urban system.
SMALL AND MEDIUM SIZE ENTERPRISES (SMEs)
It is worth stopping for a moment in order to consider exactly what is meant by an SME. Each research team and/or government body appears to have its own definition, and to date there has been no concerted attempt at standardisation. There are reasons for this. On the one hand, the definition is often dependent on the nature of the data gathered. If one source has data for companies in the 100 to 250 employee group and another for companies in the 100 to 199 employee group, there will necessarily be discrepancies in their definitions. On the other hand, the criteria used for defining an establishment may be different. The Quebec Ministry for Industry and Commerce, for instance, defines small goods producing companies as those employing less than 49 people and possessing less than $3,000,000 in assets. Small non-goods producing companies are those with less than 49 employees and less than $2,000,000 in assets (Groupe d'Analyse sur les PME et les Regions (GAPMER) 1994). Industry Canada (1987), for its part, defines small companies as those employing fewer than 100 people in the manufacturing sector and fewer than 50 people in other sectors. Medium size enterprises are variously defined as those employing between 50 and 199 people with between $3 million and $12 million of assets (GAPMER 1994), between 50 and 500 people (Industry Canada 1994), or as those companies of between about 10 to 500 employees with turnover below a certain threshold (French Government, as reported in d'Ambroise 1989).
D'Amboise (1989), reviews the various possible criteria of identification, and separates them out into quantitative and qualitative measures. Quantitative measures are measures such as employment, level of production, turnover, salaries, assets etc. Qualitative ones can include market penetration, geographic extent of market, level of autonomy and management practices. Further complications arise when the distinction has to be made between companies (which can include many branch plants), physical establishments, administrative (tax or payroll) establishments and so on. In addition, drawing the line between autonomous and branch establishments is no simple matter, since the distinction not only depends upon legal ownership arrangements and share holding but also on management arrangements and internal organisation. Although this distinction is important on a micro level, and can affect results when larger firms are considered, Picot et al. (1994) conclude that the distinction between establishments and companies (defined as legal entities in the taxation system) is not important when smaller businesses are being analysed since most are single establishment companies. In any case, it is not the purpose of this paper to explore in detail these possible definitions: rather, it is important to point out from the outset that without a well defined `target', regional policies aimed at promoting small enterprise necessarily embody a large element of uncertainty.
Despite these many different approaches, a fairly pragmatic approach is usually adopted, and while some additional criteria may be used in certain industries, SMEs are primarily defined in terms of employment. This leads to another problem linked with the definition and analysis of SMEs. Industry Canada's (1987) definition includes all but 2000 of the 2.1 million businesses in Canada. Is it really meaningful to analyse this group as a whole? Of course it is helpful to know that a disproportionate number of jobs are created by companies of less than 100 employees, which constituted 93.3% of all US establishments in the early 1980's (Birch 1987). But assuming their job creation role to be proven, not much has been learned from a policy perspective in terms of targeting intervention. For this reason, our empirical analysis looks not only at cumulative size classes (for example, all establishments of less than 100 employees) but also at discrete size classes (for example, all establishments of between 50 and 99 employees). Failure to do so can lead to erroneous conclusions and ineffective policies. (3)
Review of Recent Evidence Concerning the Job Creation Role of SMEs
As outlined above, Birch's initial research did not go unchallenged and only in 1987 did a study come out reconciling the contrasting conclusions of Birch and the Small Business Administration. Storey and Johnson (1987) determined that over the 1978-1980 period, 50% of new jobs were created by establishments of below 100 employees, which accounted for only 38% of total employment in the USA.
Birch (1987) extended his original analysis to cover a longer period, and reported that from 1969 to 1976 enterprises of between 0-19 employees created 66% of net employment growth in the USA, between 1977-1981 61% and between 1981-1985, 88%. In Canada the equivalent figure for the 1974 to 1982 period is reported as 55%. As an indication, enterprises in the size category accounted for 24.1% of employment in the US in 1980 and 26.7% in Canada in 1974.
P.A. Julien (1987) reviews the evidence available at the time concerning employment growth in SMEs. He reports evidence from Japan between 1971 and 1977, which shows that manufacturing firms of below 100 employees created the most jobs over the period. Similar figures are reported for the industrial sector in the British Midlands (1968 to 1975) and in France (1972 to 1984), and he concludes that since approximately 1970 small firms have created a disproportionate number of jobs.
Storey (1988) presents data covering 12 OECD countries over a period ranging from 1970 to 1983. The figures show the share of employment accounted for by small (under 20 employees) and large (over 500 employees) manufacturing firms. In no country does the proportion of employment in small companies decline, and in only one does the proportion of employment in large companies rise. In nine of the twelve countries there is either a rise in the proportion of employment in small companies, a decline in the proportion in large, or both combined. While the difficulty of international comparisons is acknowledged, the evidence appears to indicate a rising importance of small companies in terms of employment over the 1970's and early `80's, at least in the manufacturing sector.
These results, in particular Birch's, have come under close scrutiny, principally from Davis, Haltiwanger and Schuh (1996). Indeed, they report that small firms in the USA are not creating a disproportionate number of jobs and set out a number of methodological criticisms of Birch's work. These criticisms will briefly be summarised below.
One of the first criticisms is the fact the data used by Birch (the Dun and Bradstreet files) do not cover all firms in the USA, do not provide very accurate employment counts, and do not permit a clear identification of all firm births and deaths. Another criticism is that a number of statistical phenomena are not accounted for if the rate of employment change is analysed. First of all, the fact that firms may fluctuate around a mean level of employment means that the large firm sector will contain a disproportionate number of firms which are above their long term employment level (and are thus likely to shed jobs) and that the small firm sector will contain a disproportionate number of firms which are below their long term employment level (and are thus likely to grow). (4) Second, the boundary conditions are such that random employment fluctuation will cause the small firm class (the lower boundary of which is zero) to display positive net job growth because the lower possible growth rates are truncated (Baldwin and Picot 1994). Finally, the rate of net job change does not take into account the number of jobs created: high growth rates in small firms may still produce lower absolute employment gains than low growth rates in large firms.
A final but important criticism is that the actual measurement of firm size has an effect on the outcome: indeed, whether a firm's size is taken to be that at the start of the period of analysis, an average of the size across the whole period, or an average over two or more years either preceding or going into the analysis period, determines the classification of a firm into a given size category and hence has an important effect on the outcome in terms of employment growth by size class.
Baldwin and Picot (1994) take the above considerations into account and analyse job creation in the Canadian manufacturing sector, while Picot et al. (1994) do so for the entire Canadian economy. For the economy as a whole, it is found that, depending on the sizing definition used, employment in firms with under 20 employees grew at an average rate of 3.3% to 8.1% per year between 1978 and 1992, and that the equivalent …
Questia, a part of Gale, Cengage Learning. www.questia.com
Publication information: Article title: Establishment Size and Employment Growth in the Canadian Urban System 1991-1994: An Exploratory Analysis. Contributors: Shearmur, Richard G. - Author, J. Coffey, William - Author. Journal title: Canadian Journal of Regional Science. Volume: 19. Issue: 3 Publication date: Autumn 1998. Page number: 303. © 2008 Canadian Journal of Regional Science. COPYRIGHT 1998 Gale Group.
This material is protected by copyright and, with the exception of fair use, may not be further copied, distributed or transmitted in any form or by any means.