1The towns for which proportionality in enterprise development was investigated in this study. http://sajs.co.za/index.php/SAJS/article/downloadSuppFile/588/3665
Determination of enterprise structures The rapid method developed by Toerien and Seaman 8was used to determine the formal enterprise structures of towns. The method is based on listings of businesses in the telephone directories for the towns selected (Table 1). Enterprises were counted and categorised using 19 major business sectors (Table 2), which included economic drivers as well as service providers. 8The economic drivers are the principal business sectors that bring money into towns whilst the service sectors are probably more important in circulating money in towns than in bringing new money in. 8
2The classification of enterprises into business sectors. http://sajs.co.za/index.php/SAJS/article/downloadSuppFile/588/3666
Data analysis Pearson correlations were performed to determine proportionalities. Microsoft Excel software was used to perform the analyses and a statistically significant correlation (p < 0.05) resulted in rejection of the null hypothesis. Proportionality and the number of residents in towns The first hypothesis was tested by calculating the Pearson correlation coefficient between the populations of towns (independent variables) and the enterprise numbers per town (dependent variables) for a subgroup of South African towns. A subgroup of towns was used because formal population numbers for towns are not readily available. The exercise was complicated by a number of issues, (1) the resident numbers for South African towns are not readily available from the most recent formal census data because the latter was focused on local authorities that often include more than one town; (2) little separation is made between urban and rural (hinterland) data; and (3) the last census was done in 2001, nearly a decade ago. Consequently, estimates of town populations for 2006 reported by Hans Fransen, a well-known town planner, in his book on old Cape towns14 were used for towns that were included both in Fransen’s book and in the database on which this contribution is based (n = 35). There was a strong proportionality between population numbers per town and the number of enterprises per town as shown by a highly significant correlation coefficient (r = 0.92, p < 0.01; Figure 1). The first hypothesis was therefore accepted, a conclusion confirmed by an additional calculation of the correlation coefficient between the 2004 urban population numbers presented by Nel and Hill12 for 10 Eastern Cape towns (Cradock, Aberdeen, Jansenville, Steytlerville, Graaff-Reinet, Middelburg, Willowmore, Hofmeyr, Steynsburg and Venterstad) and enterprise numbers recorded for these towns in this study (Table 1). A statistically significant (p < 0.01) relationship with r = 0.99 was recorded: Enterprise number in town = 0.0106 (urban population number) – 21.8. Proportionality and the age of towns To test the second hypothesis, a Pearson correlation coefficient was calculated between the age of the selected towns and enterprise numbers. The founding dates of towns were found in either Fransen 14, Erasmus 15, or, if these did not provide the information, Internet searches. The only town for which a founding date could not be established was Klipplaat, which was excluded from the analysis. The correlation coefficient (r = 0.057, n = 124) was not statistically significant (p > 0.05); consequently the null hypothesis was accepted. Town age is therefore not an important general factor in determining the size of enterprise assemblages of South African towns, except for towns with similar origins. 9Proportionality and enterprise structures To examine the third hypothesis, Pearson correlation coefficients and regression equations were calculated between the total number of enterprises per town (independent variable) and the number of sector enterprises per town (dependent variables) for each of the 19 different business sectors selected (Table 2). Because the magnitude of the enterprise assemblages of towns varied from 8 for a small village such as Norvalspont, to more than 1800 for a town such as Welkom, care had to be exercised to prevent the use of spurious correlations (and spurious interpretations) resulting from the size differentials of enterprise structures of towns. Knowing that smaller towns have fewer service functions1 it was also necessary to ask if observed proportionalities extended over the whole range of enterprise assemblages in the selected town group. To examine these questions the selected group of towns was divided into three approximately equally sized subgroups: 40 large (> 130 enterprises per town), 38 medium (between 50 and 130 enterprises per town) and 47 small (fewer than 50 enterprises per town) towns. Just as for the total group of towns, Pearson correlation coefficients and regression equations were calculated and tested for each of the subgroups. Significant correlations (p < 0.05) were registered between the total number of enterprises per town (independent variable) and the number of sector enterprises of all 19 different business sectors (dependent variables) for the total town group (Tables 3, 4 and 5). Some of the regression equations explained almost all of the variance in sector enterprise numbers and others only a limited amount. Strong proportionalities were observed within the business sectors of large towns whilst fewer proportionalities were observed in smaller towns.
3Correlation coefficients and regression equations between total enterprise numbers (independent variables) and sector enterprise numbers (dependent variables) for ‘economic driver’ business sectors of groups of South African towns. http://sajs.co.za/index.php/SAJS/article/downloadSuppFile/588/3667
4Correlation coefficients and regression equations between total enterprise numbers (independent variables) and sector enterprise numbers (dependent variables) for a number of service business sectors of groups of South African towns. http://sajs.co.za/index.php/SAJS/article/downloadSuppFile/588/3668
5Correlation coefficients established and variances explained by linear relationships between total enterprise numbers (independent variables) and sector enterprise numbers (dependent variables) for the rest of the service business sectors of selected South African towns. http://sajs.co.za/index.php/SAJS/article/downloadSuppFile/588/3669
Three broad groups of business sectors were identified (Table 6), (1) sectors such as agricultural products and services, and the processing and the factory sector, where only about 25% – 30% of the variance was explained by the regression equations, (2) sectors such as mining, tourism and hospitality, news and advertising, and real estate services, where 50% – 60% of the variance was explained by the regression equations and (3) sectors such as construction, engineering and technical services, financial services, telecommunications services, trade, vehicle, general, professional, personal, health services, and transport and earthworks, where in excess of 75% (and sometimes more than 90%) of the variance was explained by the regression equations. Proportionality occurs in different business sectors of South African towns and comes in more than one guise.
6Variance (%) in sector numbers explained by group regression equations. http://sajs.co.za/index.php/SAJS/article/downloadSuppFile/588/3670
Sector proportionalities: Economic drivers Toerien and Seaman 8identified six business sectors as the economic drivers of South African towns (Table 2). With the exception of the construction sector, these sectors all had regression equations that explain a fairly limited part of the variance of sector enterprise numbers (Table 6). This phenomenon required further consideration because of its potential impact on LED planning. Many South African towns were founded to meet the spiritual needs of farmers and their families who had settled in inland areas (the so-called ‘church towns’ 14) but soon also catered for the commercial needs of farmers. The majority of South African towns, therefore, started with strong initial links to agriculture. Thus, one would expect that the enterprise structures of especially smaller rural towns should be strongly influenced by the business sector that caters for the agricultural client (i.e. the agricultural products and services sector). Although the data reveal statistically significant correlations and thus proportionality (Table 3), only limited amounts of the variances in enterprise numbers of this sector are explained by the regression equations (Table 3). Some small towns are very strong in this sector, others are very weak and the historically strong link to agriculture now appears to be tenuous. These findings support the contention that hinterlands are now less important than they were in the past for most rural towns. 12 Agriculture should not be readily positioned as the ‘saviour economic activity’ for rural local authorities struggling with LED; yet, this approach is precisely the chosen rural development strategy of the provincial government of the Western Cape. 16Tourism has been promoted as an economic activity that can contribute significantly to rural economic development in South Africa. 17The number of tourism and hospitality sector enterprises in towns was significantly correlated with the total number of enterprises per town, but only 58% of the variance in the enterprise numbers of this sector was explained by the regression equation (Table 6). Larger towns are nevertheless likely to have reasonably strong tourism and hospitality sectors. In the medium and small town groups much less of the variance in the sector was explained by the regression equations (Table 6). Some towns in these groups have strong tourism and hospitality sectors but others have very weak tourism and hospitality sectors. Successful positioning of towns in the tourism and hospitality sector is neither an automatic nor an easy process. Value addition has been promoted as a means of adding economic value in South African towns. 18Two business sectors that add value to either local and/or external primary and/or intermediary materials are the processing and factory sectors. 8Success in these sectors had little association with the size of the enterprise assemblages of South African towns (Tables 3 and 6) and smaller towns often had more processors and/or factories than did larger towns. Success in these sectors depends on aspects such as ‘special’ entrepreneurship, talented individuals and management according to the key success factors of a specific business model. 19In this study, relatively small towns were quite successful in the processing sector, suggesting that it is not beyond the capabilities of small towns to achieve success. The mining sector needs special consideration. The statistically significant correlations (and hence proportionalities) registered between the numbers of mining-related enterprises and the total number of enterprises for the total town group, as well as for the large town groups (Table 3), could be spurious and might have been caused by the impact on the calculations of the large mining sector of Welkom. However, a large mining sector is not a precondition for significant business development, because some other large towns in this study did not have significant mining sectors. The mining sector also did not have a major impact on the enterprise assemblages of medium and small towns and there were no proportionalities (Table 3). The only economic driver sector that showed significant proportionality throughout is the construction industry (Table 3). This proportionality only really applied to large towns, because although there was a positive correlation between the number of construction enterprises and the total number of enterprises in the medium and small towns, the degree to which the variance could be explained was low and towns of equal size can differ widely in the number of their construction enterprises. The fact that large towns have proportionally larger construction sectors and small towns sometimes have few, if any, suggests that some, if not all, construction activities in smaller towns might be provided by suppliers from the informal sector. Sector proportionalities: Service sectors Proportionalities in the service sectors (Table 2) were much higher than in the economic driver sectors (with the exception of the construction sector) (Tables 4 and 5) and more of the variance was explained by the regression equations (Table 6). The outstanding example of service sector proportionality that extends right through all of the size ranges of towns is the trade sector (Table 6). Amongst the medium and small towns there were larger variances than what was encountered for the large town group. Nevertheless, proportionality remained strong over all of the groups (Tables 4 and 5). Other service sectors with reasonably strong extended proportionalities were the financial services, legal services, general services, personal services, health services, and the transport and earthworks sectors (Tables 4 and 5). The engineering and technical services, telecommunication services, news and advertising services, vehicle, professional services, and real estate services sectors showed strong proportionalities as far as the large towns were concerned (Tables 4 and 5). These proportionalities apparently broke down for smaller towns. Just as in the construction industry, shifts across the interface between the formal and informal economies might be the cause. In summary, whilst the economic driver sectors, with the exception of the construction industry, exhibited limited proportionalities (Table 3), the service sectors had more pronounced proportionalities, especially for large towns (Tables 4 and 5). Proportionality and thresholds Van der Merwe and Nel 20provided estimates of threshold numbers for a variety of different types of service suppliers. Using total population and total enterprise numbers to calculate threshold values assumes a linear relationship between the population and the number of enterprises. This study confirmed the existence of a linear relationship (Figure 1). A further question arose as to whether linear relationships also applied to sector enterprises. Some of the towns in this study were towns for which Fransen 13provided population estimates (Table 7), which allowed investigation of the relationship for sector enterprises. Regression equations were calculated for town population numbers (independent variable) and sector enterprise numbers for those business sectors with extended proportionalities (Table 8). Because some of these business sectors did not have strong proportionalities in the medium and small town groups, distortions of the regression equations could have resulted. To check if this had happened the above calculations were repeated (Table 9) for those large towns (more than 130 enterprises) included in Table 7. Comparison of Tables 8 and 9 indicates that, if distortion was introduced, it was not severe.
7Towns analysed in this study that were included in Fransens13 estimates of population sizes. http://sajs.co.za/index.php/SAJS/article/downloadSuppFile/588/3671
8Correlation coefficients, regression equations, variances explained and carrying capacities derived from comparisons of population estimates and enterprise numbers of business sectors with extended proportionalities for all towns in Table 7 (n = 35). http://sajs.co.za/index.php/SAJS/article/downloadSuppFile/588/3672
9Correlation coefficients, regression equations, variances explained and carrying capacities derived from comparisons of population estimates and enterprise numbers of business sectors with extended proportionalities for large towns (> 130 enterprises) in Table 7 (n = 14). http://sajs.co.za/index.php/SAJS/article/downloadSuppFile/588/3673
The precision of the above regression equations is confounded by the reality that population estimates of town populations rather than hard census data had to be used. The 2011 South African census should provide better estimates of town populations for future analyses. DiscussionEnterprise proportionality is worthwhile investigating for a number of reasons. This study augmented the original observations about threshold populations1 by revealing that there are currently still strong proportionalities between the total number of enterprises per town and the number of sector enterprises per town for most of the services sectors presented in Table 2. Regression equations quantified these relationships (Tables 3, 4 and 5) and much of the variances of particular sector enterprise numbers were explained (Table 6). Proportionalities were also linked to the population numbers of towns in general (Table 8) and especially to those of the large towns (Table 9). A comprehensive literature search revealed that little has been reported, internationally as well as locally, on the reasons why proportionalities are present and could be expected in the enterprise structures of towns. For instance, a recent report on the regeneration of small towns in South Africa paid little attention to this issue. 21The exception is provided by Toerien and Seaman 9who likened enterprise development in South African rural towns to islands rising above the sea and over time being colonised by all kinds of living organisms. As new space becomes available on a natural island, some immigrant organisms will use it. 10Toerien and Seaman 9showed that the Species Equilibrium Model of MacArthur and Wilson 10can be used to describe the dynamics of enterprise development in South African towns. They suggested that when a new town is founded, entrepreneurial space becomes available and is utilised by entrepreneurs who, similar to their organism counterparts, have to move across a ‘sea’ to reach the opportunities. Not all entrepreneurs are successful in the competitive business environment and some enterprises go under. Over time equilibrium is reached between the rate of development of new enterprises and the rate of disappearance of existing enterprises and this phenomenon plays a major role in defining enterprise proportionalities. In addition, issues such as town size, population size and town age also played a role in determining the proportionalities of the 12 Karoo towns. 9The question as to why there are such marked differences in proportionalities between most of the economic driver sectors and the service sectors deserves attention. In particular, the difference(s) between the utilisation of entrepreneurial space in the trade sector (Table 4) in contrast to, for example, the processing sector (Table 3), needs clarification. A specific question can be posed: why is there such a strong correlation in the trade sector over 125 towns of different sizes, ages, physical localities and origins, and why is it not the case for the processing sector? Considering Beinhocker’s 22discussion of competition between business plans, and hence between enterprises, different proportionalities in the two sectors probably suggest that entrepreneurial space is used differently by the entrepreneurs of the two sectors. In the trade sector, entrepreneurs are able to ‘see’ business opportunities easily and exploit them. As a consequence, when one business disappears as a result of competition it is rapidly replaced, usually by the actions of a different entrepreneur who is willing to take on the risks involved. This phenomenon represents ‘run-of-the-mill’ entrepreneurship with entrepreneurs that can ‘see’ and react to ‘standard’ business opportunities with perceived limited risks. In contrast to the above, value addition in the processing and factory sectors is not pursued effectively in South Africa and small towns may have more of these enterprises than very large towns (Table 3). Obviously, available entrepreneurial space is not being utilised fully, perhaps because it is much more difficult to ‘see’ these business opportunities and they probably are more risky and require significantly higher capital inputs. Success in these sectors is dependent on ‘special’ entrepreneurs. The basic differences between enterprises founded upon ‘run-of-the-mill’ and ‘special’ entrepreneurship are fundamental. ‘Run-of-the-mill’ enterprises are dependent on, and limited by, local demand and if there is only a small demand, the entrepreneurial space is small. In contrast, ’special’ enterprises have much larger markets because their products and/or services are exportable, or in other words, they are potential global players and have an unlimited scope. The implications for rural development and regeneration of these two broad entrepreneurial types are considerable. For instance, the results (Table 4) suggest very strongly that the trade sector offers little scope to expand local economies because the entrepreneurial space in the trade sector is already fully exploited and there is little chance to create additional opportunities for emergent entrepreneurs. When a new enterprise is started in a sector that is already congested and in balance with the market demand of this sector, it will be subject to strong competition. If it manages to scrape out a place in the sector, it will do so at the expense of another enterprise that already operates in that sector. A system of ‘musical chairs’ will ensue where one enterprise is merely replaced by another but the total ‘game’ has not grown. To expand the economies of South African towns, the money spent in the economies of these towns must be grown. This challenge is huge. Nel and Hill 12outlined some of the complicating factors, for example, agriculture being under pressure resulting in a loss of farmers and farm workers as clients of towns’ enterprises; better roads and transport infrastructure resulting in rural and town residents making their purchases in the larger towns of a region; the loss of personnel (and their wages) of government departments and organisations such as the South African Railways, and the loss of schools. Sharp contrasts exist in LED practice between large cities with their focus on competitiveness, secondary cities with a focus on economic readjustment and small towns in which LED practice is limited or laggard. 17Given these challenges, it is clear that development strategies must focus on how the flow of money into towns can be enhanced. Success in the mining sector depends on the presence of mineral resources, which cannot be planned. Therefore, growing the processing, factory, tourism and hospitality, and construction sectors must be a serious consideration in most strategic LED plans for towns because these sectors offer opportunities to increase the flow of money into towns. Several studies have indicated that many municipalities lack an understanding of their economic spaces. 17,21 The hypothesis outlined here, of two broad kinds of entrepreneurship in South African towns, offers a new way of thinking about rural development and job creation. It needs to be investigated further and a better understanding of the proportionality phenomena is a key to such an endeavour. Acknowledgements We gratefully acknowledge the financial support of the Centre for Environmental Management, University of the Free State; the library support of Annamarie du Preez and Estie Pretorius; and the analytical support of Marie Toerien. Competing interests We declare that we have no financial or personal relationships which may have inappropriately influenced us in writing this article. Authors’ contributions D.F.T. was responsible for the conceptualisation of the basic ideas, the analyses of the enterprise assemblages of towns, the interpretation of the results and writing the manuscript. M.T.S. refined some concepts and provided advice from an ecological perspective. 1.Van der MerweIJNelAThe city and its environment: A study in settlement geography2.DaviesRJCookGPReappraisal of the South African urban hierarchy1968501161323.FujitaMKrugmanPVenablesAJThe spatial economy: Cities, regions, and international trade2001144.RomerPMEconomic growthHenderson DR5.FujitaMKrugmanPVenablesAJhe spatial economy: Cities, regions, and international trade6.FloridaRThe rise of the creative class7.BeinhockerEDThe origin of wealth: Evolution, complexity, and the radical remaking of economics8.http://dx.doi.org/10.4102/sajs.v106i5/6.1829.ToerienDFSeamanMTEvidence of island effects in South African enterprise ecosystems20936099295093910. SchoenerTWThe MacArthur-Wilson equilibrium model: A chronicle of what it said and how it was tested11.ZumaJWe remain committed to taking the economic transformation forward12.http://dx.doi.org/10.1016/j.jaridenv.2008.07.01513. FransenHOld towns and villages of the Cape14.FransenHOld towns and villages of the Cape15.ErasmusBPJOn the road in South Africa16.Van RensburgGRural areas are not no-man’s-land17.RogersonCMStrategic review of local economic development in South Africa: Final report submitted to minister Sicelo Shiceka (dplg)18.ToerienDFTaming Janus: Technology, business strategy and local economic development19.ToerienDFTaming Janus: Technology, business strategy and local economic development20. Van der MerweIJNelAThe city and its environment: A study in settlement geography21. WesselsJDevelopment of a small town regeneration strategy: Final Report22.BeinhockerEDThe origin of wealth: Evolution, complexity, and the radical remaking of economics