10
Geoforum,Vol. 14,No l,pp.?S-54.19X3 Printed in Great Britain 0016~71X5/83/01IX145-I0 $03 OMI 0 I983 Pcr&mmn Prrs Ltd Observations on the Relevance of the Industrial-Urban Hypothesis for Rural Development DANIEL TODD,* Manitoba, Canada Abstract: Rural development tends to be viewed as a facet of urban development in the sense that rural areas are expected to be recipients of positive urban spill-overs. To that end, two concepts have been formulated: growth centre theory and the Industrial-Urban Hypothesis (IUH). The former has been beset by con- ceptual problems whereas the simpler IUH has not been tested in a Canadian context. This paper tests the basic IUH proposition that the general level of rural development is an inverse function of urban accessibility. Findings suggest that the IUH is more attuned to the impacts of smaller centres rather than the region’s metropolitan centre. However, introduction of an exogenous element based on ethnic variations in rural settlements indicate how socializing factors work to modify the IUH concept. Introduction In spite of some strong reservations to the contrary, rural development is generally perceived to be predicated on the effectiveness of urban development.’ This occurs as a result of positive development impulses - spread-effects in Myrdal’s parlance - disseminated by the urban centres which are assumed to be responsible for rural ‘modernization’. Consequently, an understanding of the forces responsible for rural development reduces, in the main, to a concern for the patterns of spatial diffusion of development influences from urban centres. Indeed, one of the main claims for relevance of growth centres is that such induced planning instruments serve their rural tributary zones in precisely this manner, that is, they act as the prime agents for initiating positive change (social as well as economic) and through that, the well-being of rural residents. The aim of this paper is to review a longstanding conceptual framework for urban-based rural development and establish a case study from the *Department of Geography, University of Manitoba, Winnipeg, Canada. prairie province of Manitoba in order to gain insights into its validity. The concept in question is the Industrial-Urban Hypothesis, so called because it maintains, a priori, that industrialization and urbanism are inseparable and, as such, designates urban centres as the keys to economic development. In addition, the Industrial-Urban Hypothesis (hereafter IUH) goes a step further in presuming that the urban centre is the catalyst for rural development. In a nutshell, the concept holds that the degree of rural transformation is largely depen- dent upon ease of access to an urban centre. Those localities within reasonable proximity to a centre will be most subject to the development spin-offs issuing forth from the centre whereas isolated rural districts will not be penetrated by such impulses and will experience a corresponding stultification of development. Evidently, there is much in common between the IUH and growth centre thinking, but the main difference is that the genesis of the former was couched specifically in terms of rural develop- ment with little regard for the mechanisms which gave pre-eminence to the urban centre. In marked contrast, growth centre theory places its main emphasis on urban mechanisms, relegating almost to an aside consideration of the wherewithal for transference of spread-effects to rural areas. 45

Observations on the relevance of the industrial-Urban hypothesis for rural development

Embed Size (px)

Citation preview

Page 1: Observations on the relevance of the industrial-Urban hypothesis for rural development

Geoforum,Vol. 14,No l,pp.?S-54.19X3 Printed in Great Britain

0016~71X5/83/01IX145-I0 $03 OMI 0 I983 Pcr&mmn Prrs Ltd

Observations on the Relevance of the Industrial-Urban Hypothesis for Rural

Development

DANIEL TODD,* Manitoba, Canada

Abstract: Rural development tends to be viewed as a facet of urban development in the sense that rural areas are expected to be recipients of positive urban spill-overs. To that end, two concepts have been formulated: growth centre theory and the Industrial-Urban Hypothesis (IUH). The former has been beset by con- ceptual problems whereas the simpler IUH has not been tested in a Canadian context. This paper tests the basic IUH proposition that the general level of rural development is an inverse function of urban accessibility. Findings suggest that the IUH is more attuned to the impacts of smaller centres rather than the region’s metropolitan centre. However, introduction of an exogenous element based on ethnic variations in rural settlements indicate how socializing factors work to modify the IUH concept.

Introduction

In spite of some strong reservations to the contrary, rural development is generally perceived to be predicated on the effectiveness of urban development.’ This occurs as a result of positive development impulses - spread-effects in Myrdal’s parlance - disseminated by the urban centres which are assumed to be responsible for rural ‘modernization’. Consequently, an understanding of the forces responsible for rural development reduces, in the main, to a concern for the patterns of spatial diffusion of development influences from urban centres. Indeed, one of the main claims for relevance of growth centres is that such induced planning instruments serve their rural tributary zones in precisely this manner, that is, they act as the prime agents for initiating positive change (social as well as economic) and through that, the well-being of rural residents.

The aim of this paper is to review a longstanding conceptual framework for urban-based rural development and establish a case study from the

*Department of Geography, University of Manitoba, Winnipeg, Canada.

prairie province of Manitoba in order to gain insights into its validity. The concept in question is the Industrial-Urban Hypothesis, so called because it maintains, a priori, that industrialization and urbanism are inseparable and, as such, designates urban centres as the keys to economic development. In addition, the Industrial-Urban Hypothesis (hereafter IUH) goes a step further in presuming that the urban centre is the catalyst for rural development. In a nutshell, the concept holds that the degree of rural transformation is largely depen- dent upon ease of access to an urban centre. Those localities within reasonable proximity to a centre will be most subject to the development spin-offs issuing forth from the centre whereas isolated rural districts will not be penetrated by such impulses and will experience a corresponding stultification of development. Evidently, there is much in common between the IUH and growth centre thinking, but the main difference is that the genesis of the former was couched specifically in terms of rural develop- ment with little regard for the mechanisms which gave pre-eminence to the urban centre. In marked contrast, growth centre theory places its main emphasis on urban mechanisms, relegating almost to an aside consideration of the wherewithal for transference of spread-effects to rural areas.

45

Page 2: Observations on the relevance of the industrial-Urban hypothesis for rural development

46

Because the onus of this paper is on rural develop- ment, attention first needs to be directed to the explicit rural-oriented propositions of the IUH before returning to the rural implications of growth centre notions.

The Essence of the Industrial-Urban Hypothesis

The IUH idea derives directly from the Location Matrix model promulgated by SCHULTZ (1953). This model explicitly links rural, and in particular, agricultural prosperity with spatial proximity to an urban centre. This is so because:

“The existing economic organization works best at or near the centre of the particular matrix of economic development and it also works best in those parts of agriculture which are situated favourably in relation to such a centre; and it works less satisfactory in those parts of agriculture which are situated at the periphery of such a matrix” (SCHULTZ, 1953: 147).

In according pride of place to the urban centre, Schultz is assuming that it serves to eliminate supply and market imperfections. Thus, the centre acts to allocate efficiently labour, capital and land supply on the one hand, and also to absorb rural products on the other. Any farmer who comes under the sway of urban organization will profit both from greater factor returns (e.g. higher value land, benefits from increased capitalization, and the greater productivity of better-trained labour) and higher sales as a result of access to the centre’s larger and better organized produce markets.2 In turn, higher sales volumes feedback to enable the farmer to further his capital intensification and pay higher wages for more qualified labour; the two combining to ensure the self-sustenance of rural development. Certainly, there is evidence to sug- gest that these assertions are plausible. Summariz- ing studies undertaken in the Tennessee Valley and Piedmont regions of the United States, NICHOLLS (1961) accedes very strongly to the view that urban factor markets have contributed to the efficiency of nearby agriculture. Moreover, at the individual fac- tor level, other studies have suggested that farm labour supply is closely related to urban manpower markets (SMIT, 1979) and that the degree of farm capitalization is a function of urban orientation (GREGOR, 1979).

Although dealt with only tentatively, the original Schultz concept assumed that urban strength stem-

GeoforumiVolume 14 Number 111983

med from the concentration of industrial activities in the centres. This proposition has become the cornerstone of the IUH in the sense that industrial urbanism is advocated as the main agent in fostering not only the static efficiency of agriculture, but also such diverse dynamic factors as the rate of innov- ation diffusion and the progress which follows from rises in the level of rural education (KATZMAN, 1974). Recently, the servicing function of centres, long recognized in central place theory, has also been formally grafted onto the IUH concept (TAURIAINEN and YOUNG, 1976). The out- come is that the IUH now accommodates an urban role designed to respond to the comprehensive needs of the farmers for it no longer requires the centre to be primarily an industrial cluster. Two obvious implications stem from this revised IUH: firstly, the hierarchical nature of servicing consoli- dates the relationship between regulator (centre) and regulated (rural area) and secondly, smaller urban centres without significant industrial appen- dages (analogous to relatively low-order central places) can, nonetheless, be conceived as possible leading agents in rural development.

The stage is now set for summarizing the variety of means whereby the urban centre acts to influence development of adjoining rural areas. Among the more likely outcomes posited by modified IUH thinking are the following five.

1. A declining average rural income gradient away from the centre as increasing isolation denies the rural resident access to urban labour-markets and removes pressure on farmers to increase wages for agricultural workers. 2. Similarly, average farm incomes may decline with distance from the centre because those farm families within reasonable commuting range may supplement their incomes with urban work. Fur- thermore, access to markets through better trans- port facilities are likely to boost the profits of far- mers living in the vicinity of the centre. 3. By the same token, access to urban educational and financial institutions may improve the profit- ability of the neighbouring farm through greater labour productivity (which is a natural corollary of more skillful workers, including the farmer himself) and increased capitalization (for which the farmer can use his high location rent as collateral). 4. Further, the existence of more efficient produc- ers in urban centres may benefit those rural businesses with easy access to the cities. From the point of view of agriculture, this could mean

Page 3: Observations on the relevance of the industrial-Urban hypothesis for rural development

GeoforumiVolume 14 Number 111983 47

reduced costs for those farmers who can readily tap

urban suppliers. Again, a distance-decay function is implicit in terms of such advantage. 5. The conception of development as an inverse function of distance from urban centre is not inviol- able. Small communities at some distance from the major urban centre may possibly gain from iso- lation. This would occur if isolation served to impede competition from city producers and enabled small businesses to prosper. In addition, the relative infrequency of contracts with the city on the part of rural residents may work to ensure that there are less ‘leakages’ out of local income circu- lation. In turn, a robust multiplier effect would be conducive to more local job-creation.

While debate continues on the validity of the detailed manifestations of these urban spill-overs, there appears to be little doubt that rural districts are in some way beneficiaries of urban develop- ment, if only that taking place in large centres. Insofar as income operates as a general indicator of development, it has been noted in the United States that during the 1960s:

“Median family income change was highly associated with accessibility to, and growth of, metropolitan cen- tres. Spread effects from the metropolis have clearly been a key factor in raising non-metropolitan welfare levels, both inside and outside metropolitan commut- ing fields” (LAMB, 1975, p. 183).

The Growth Centre Affinity

The allusion to spread-effects in the above quo- tation hints at the linkage between the IUH and growth centre theory. Notwithstanding different terminologies, the positive urban spill-overs deemed essential for rural development by IUH proponents are basically the same as the spread- effects of growth centre theorists. Indeed, growth centre theory places great stress on the conditions conducive to dissemination of such development impulses from planned urban nodes. To this end, much effort is focused on identification of the econ- omic structure that needs to be nurtured in growth centres before spread-effects can be guaranteed. Unhappily, it is at this juncture that theory becomes divorced from reality because the onset of spread- effects has been shown not to be a fait accompli.

Right from the beginning of thinking along growth

centre lines, controversy has raged as to whether positive or negative spill-overs emanate from urban centres.3 The fundamental dichotomy between Hirschman’s opting for optimism and Myrdal for pessimism bears witness to the inadequacy of the growth centre as a prescriptive device for compre- hensive development. Since that time, attempts to clarify the situation through empirical analyses have only added further confusion. On one side are students who have demonstrated the existence of positive development-response surfaces in rural areas abutting on growth centres in regions as far apart as Brittany, East Anglia and Malaysia (MOSELEY, 1973; ROBINSON and SALIH, 1971). On the other side are studies which dismiss the prevalence of any positive spill-overs at all from growth centres (CONROY, 1973; GILBERT, 1975). Gilbert goes so far as to demonstrate the form of a negative development-response surface in hinterland areas. Of late, theorists have attempted to reconcile the conflicting evidence. RICHARD- SON (1976), for instance, has reset the model to allow for the transference of joint positive and nega- tive impulses in areas adjoining growth centres.

Despite all efforts, growth centre theory remains vague as to its treatment of urban spill-overs. In part this is due to the emphasis extended by the theory to growth-generation rather than to the mechanisms of growth-transmission. As many diffu- sion analysts have noted, much economic activity is organized according to hierarchical systems linking together urban centres of different size. Hence, transmission of multiplier effects, as well as innov- ation messages, tends to reside in inter-urban chan- nels and bypasses rural areas altogether. When it is recalled that these same activities are usually viewed as the essential ingredients for triggering growth in the planned centres, it is not altogether surprising that contagious diffusion of spread- effects fail to materialize. In emphasizing the build- up of urban-based development rather than the ramifications of such development on rural areas, growth centre theory comes close to missing the point of rural development. Therefore, as presently constituted, it cannot replace the explicit consider- ation of urban spill-overs as furnished by the IUH. At best, growth centre theory can serve to designate those urban centres with the capacity to stimulate positive spill-overs, but, at worst, it may contrive to undercut existing rural enterprise by advocating policies capable only of stimulating backwash- effects.

Page 4: Observations on the relevance of the industrial-Urban hypothesis for rural development

48 GeoforumiVolume 14 Number 111983

An Empirical Case wider array of elements at work and it is so used here.

The prairie component of the Province of Manitoba was the setting for testing the relevance of IUH thinking with respect to rural development. This region was selected because it fulfils the main criteria for potential IUH spill-overs, namely, it embraces an extensive, settled agricultural area dominated by one metropolitan centre and a few (five) much smaller secondary centres, each of which has some record of growth impacts on sur- rounding districts (TODD, 1979). In addition, the separation of urban places into two groups is useful in that it allows for an analysis of the IUH after controlling for size differences in sources of growth generation. Like other IUH studies (TAURIAINEN and YOUNG, 1976), this one is set up to compare the indices of rural development between two times; in this instance 1961 and 1971.

Obtaining development indices for comparison imposes something of a quandary. Development is a nebulous concept to begin with and deciding on its precise constituents runs the risk of applying subjec- tive values as to what is, and is not, relevant. Moreover, even though numerical summarizing procedures can measure different degrees of development, they arbitrarily equalize social vari- ation with economic variation. In short, they are useful as data-reduction techniques rather than precise evaluators of the hierarchy of forces involved in development. In this view, principal component analysis can be used to concoct a gen- eral index of development representative of a much

The variables used as input into the component analysis are drawn from the precedents established by the aforementioned IUH theorists. Accordingly, they can be divided into sub-sets indicative of capi- tal efficiency (K), labour efficiency (L), intensity of land use (P), industrialization (I), and com- mercialization (C). In total, 14 variables are involved in constructing the index: (K,) average per capita farm taxes, (K2) ratio of number of farms selling more than $10,000 to total farms, (K3) employment in finance, (L,) number of non- resident farmers, (L2) number of rural residents with a modicum of high schooling, (LX) number of rural residents with some exposure to university education, (L4) net out-migration over the previous five years, (Pi) ratio of crop land to total farm area, (PJ ratio of pasture to total farm area, (I,) employ- ment in manufacturing, (f2) employment in con- struction, (Ci) employment in trade, (CJ employ- ment in transport and (C,) employment in services. All variables are expected to contribute directly to rural development - even net out-migration which, according to neo-classical-factor equilibrating tenets, reduces the pressure of population in rural areas and so clears the way for improved labour productivity. These variables are collected for 1961 and 1971 over 114 rural municipalities (i.e. exclud- ing any places defined as ‘urban’).

The leading components for the two years are pre- sented in Table 1. Three components emerge with

Table 1. Principal components constituting development indices

Variables

K1 taxes K2 high farm sales K3 finance Lr non-resident

farmers L2 high schooling L3 university Z4 net out-migration PI crop land P2 pasture Zr manufacturing I2 construction Cr trade C, transport C, services A

1961 component loadings 1971 component loadings

1 2 3 1 2 3

0.074 0.860 4.081 0.062 0.884 a.257 0.154 0.833 a.145 0.322 0.553 0.351 0.488 0.180 Al.385 0.787 0.028 0.101

0.502 0.081 0.620 0.383 0.436 0.535 0.773 0.280 0.233 0.922 0.068 0.121 0.796 0.197 0.037 0.857 0.012 0.157 0.427 -0.193 0.697 AI.036 AI.012 0.615 0.542 0.318 AI.081 0.397 0.682 -0.144

Al.145 AI.553 a.218 0.089 -0.607 0.424 0.777 Al.215 -0.407 0.814 Al.223 AI.300 0.821 -0.290 4.239 0.798 a.248 AI.209 0.899 a.192 -0.137 0.944 Al.093 -0.093 0.868 4.168 JI.093 0.889 AI.212 Al.111 0.908 -0.155 0.140 0.939 JI.060 0.011

5.91 2.25 1.43 6.44 2.28 1.26

Page 5: Observations on the relevance of the industrial-Urban hypothesis for rural development

Geoforum/Volume 14 Number l/1983

eigenvalues in excess of one for both years; those for 1961 accounting for over 68% of variance while the 1971 components together amounting to 79% of variance. In sum, the components encompass all of the variables instrumental in rural development, although the relative contributions of variables dif- fer between components and years. Thus, in 1961 the first component had substantial loadings for services, trades, transport and construction sectors in that order, whereas the 1971 equivalent compo- nent loaded highest on trades, services, high school- ing, transport, university education and manu- facturing. Similarly, changes in emphasis were evi- dent in the second and third components. At the beginning period, the second component focused on capital efficiency, but it had veered to a combi- nation of capitalization and land use by the end period. Initial onus of the third component was on the logical interplay of out-migration and non- resident farmers, although this had been revised by 1971 to highlight an inverse relationship between high farm sales and out-migration.

Aggregation of component scores for the three dimensions provides development indices for the two years.4 Figures 1 and 2 display the patterns of composite development for 1961 and 1971. At the former time, 54 rural municipalities recorded posi- tive values for their development index, leaving the majority (60) to register negative values. By the latter time, the imbalance had worsened: 51 reg- istered positive development and 63 negative development. Indeed, the coefficient of variation in the incidence of rural development diverged from 98% in 1961 to 112% in 1971. This finding has alarming overtones inasmuch as it suggests that spa- tial disparities in development have widened over the ten year span.

The question remains as to how far the pattern of rural development has been influenced by urban accessibility in the manner posited by the IUH. Superficially, this can be gauged by noting the degree of attenuation of average development levels with distance from urban centres. Table 2

\ LEGEND

POSltlVe

m Negatave

m Urban Centre

Figure 1. Development index, 1961

Page 6: Observations on the relevance of the industrial-Urban hypothesis for rural development

50 GeoforumiVolume 14 Number 111983

Figure 2. Development index, lY71.

suggests that average development scores decline with distance from both Winnipeg and the five other regional centres in Manitoba, although this is not

Table 2. Relationship of average development scores to urban accessibility

km bands DTW* DOCt

1961 1971 I961 I’)71

l-39 0.38 0.40 1.12 1.56 40-79 0.76 0.28 0.30 0.11 80-119 0.21 0.16 4.11 -0.35

120-1.59 0.41 0.17 XI.01 -0.09 160-199 -0.42 -0.47 -0.72 -0.48 200-239 -0.25 -0.22 0.58 0.23 240-279 0.16 0.37 0.44 a.11 280-319 -0.68 -0.93 PI.45 1.21 320-3.59 -0.34 -0.50

>360 0.68 1.74

*DTW = distance to Winnipeg. tDOC = distance to other centres.

very pronounced. On the whole, however, the table does indicate that the spread of positive develop- ment is much more localized around the smaller centres that it is around Winnipeg. Interestingly, the distribution of positive development appears to have become even more concentrated in the vicinity of both Winnipeg and the regional centres by 1971. In so doing, this is corroboration of sorts for the seeming divergence of intraregional rural develop- ment over the period of study.

A more explicit account of the relationship between development and urban accessibility can be obtained from the use of regression analysis. In accord with IUH thinking, the development index is appointed as the dependent variable and the dis- tance to Winnipeg (IT), and to other centres (OC), act as the independent regulators:

Y = j30 + j3i 1nW + fl2 1nOC + Ui (1)

where Y can be separated into four components:

Page 7: Observations on the relevance of the industrial-Urban hypothesis for rural development

Geoforum/Volume 14 Number l/1983 51

1961 positive development scores, 1961 negative development scores, 1971 positive development scores and 1971 negative development scores. Ostensibly, the degree of rural development (Y) should be an inverse function of the two distance parameters, although these latter would be expected to have a weaker influence on areas of negative development (i.e. those not in receipt of spread-effects) than on those rural municipalities bestowed with positive development. Incidentally, Equation 1 is structured in semilog form to ensure intrinsic linearity for the distance relationships and also to gauge the constant absolute change in development commensurate with a given prop- ortional change in distance, that is:

exhibit zero autocorrelation (as expressed through the Durbin-Watson ‘d’ statistic), the equation for 1961 negative development is positively autocorrel- ated and the presence of covariance among the disturbance terms cannot be ruled out for the 1971 positive development regression. These findings suggest that rural areas characterized by negative development in the commencement year were spa- tially clustered, but that the incidence of such char- acteristics had become much more dispersed ten years later. By the same token, positive develop- ment was more equitably distributed across the region in the beginning than it was in 1971.

The relationships displayed by the equations imply that areas of positive development had a direct relationship with distance to Winnipeg and an inverse relationship with distance to other centres for both years. The former relationship, albeit insig- nificant, suggests that isolation from metropolitan influences may indeed be beneficial to rural areas. Perhaps this is substantiation for the notion that rural enterprises may prosper when deprived of excessive competition from metropolitan activities (e.g. manufacturers may obtain cheaper labour out- side urban labour markets). The latter relationship,

j3, = dYl(dWIW)

fi2 = dYl(dOCIOC). (2)

Table 3 indicates the urban influence on rural development for 1961 and 1971 (columns A and C). All four equations pertaining to the different variants of Y are significant and at least one of the distance parameters in each equation is also signifi- cant. However, while the equations for 1961 posi- tive development and 1971 negative development

Table 3. Regression results

A

lnD7W 0.228 (0.793) 1nDOC a.723 (2.883)***

British Mennonite

Constant 2.717 adj RZ 0.089

F 4.159** d 1.7s**

lnD7W -0.333 (1.522)* 1nDOC 0.494 (2.172)**

French Ukrainian

Constant 0.739 adj RI 0.046

F 2.386* d 1.18

Positive Development Index

1961 1971

B C

0.342 (1.09) 0.252 (0.784) -0.709 (3.869)*** -1.23 (4.088)***

0.005 (0.522) 0.034 (3.02 )*** 1.59 4.141 0.24 0.244 5.176*** 9.056*** 2.13** 1.41

Negative Developme hnt Index

1961 1971

XI.387 (1.695)* 0.532 (2.347)**” 0.004 (0.588) 0.009 (1.866)* 0.598 0.070 2.oY8* 1.31

-0.327 (1.849)** 0.352 (1.944)**

1.129 0.045 2.424* 1.56***

D

0.554 (1.396) -1.404 (4.352)*** a.007 (0.622)

0.015 (1.203) 3.762 0.253 5.211”** 1.45

-0.267 (1.602) 0.365 (2.254)*** 0.015 (2.769)*** 0.016 (3.678)*** 0.441 0.238 5.X28*** 1.64***

Bracketed values are ‘t’-statistics; number of asterisks denote level of significance (0.10, 0.05, 0.01 respectively).

Page 8: Observations on the relevance of the industrial-Urban hypothesis for rural development

52

however, is much more significant, suggesting that rural development is closely associated with the for- tunes of smaller urban centres. Indeed, a unit increase in 1nOC (equalling 40 km) will reduce the development index by 0.72 in 1961, and much more, by 1.23, in 1971. When the average development score in the vicinity of regional centres was only 1.12 (and 1.56 in 1971), this was a telling impact. Comparison of R* values for the two years (adjusted to account for differing degrees of freedom and hence to enable comparison) intimates a marked degree of improvement in level of explanation by the latter time. Evidently, this is attributable to the increasing relevance of the OC variable, which is to say, by 1971 smaller centres in Manitoba had an increased ability to stimulate rural development in areas immediately accessible to them, but at the expense of imposing negative spill-overs elsewhere.

Areas of negative development display opposite dis- tance relationships to their counterparts experienc- ing positive development. In other words, the degree of negative development is inversely assoc- iated with distance to Winnipeg and directly assoc- iated with distance to other centres. However, these findings are fully consistent with the positive development coefficients, that is, they imply that increasing distance from the metropolis reduces the incidence of negative development whereas increas- ing distance from smaller centres promotes it. In 1971, as well as 1961, a 40 km increase in distance away from Winnipeg was tantamount to boosting rural development by a score of 0.33; but the equiv- alent distance away from other centres undercut development by 0.49 in the first year and 0.35 in the second year.

Clearly, the supposition that rural development is an inverse function of accessibility to all urban cen- tres is too simple; an obvious distinction is emerging between the characteristics of smaller centre spill- overs which conform to IUH precepts and those pertaining to the metropolitan centre which do not. Yet, the levels of explanation (R*) produced by the regression equations are low and thereby hint at the prevalence of other factors in promoting rural development. In the United States, IUH students have pointed out that fundamental impediments to development may stem from social factors; in par- ticular, they may hinge on the relationship between blacks and whites in rural areas (NICHOLLS, 1961). Ethnic factors may be responsible for varia- tions in rural economic performance in Manitoba too (TODD and BRIERLEY, 1977). Accordingly,

GeoforumiVolume 14 Number l/1983

the regression equations are modified to incorpor- ate an ethnic factor:

Y, = /lo + fli 1nW + fi2 1nOC + P3B + j?4M + ui

(3)

Y, = /3<, + PI 1nW + j3,lnOC + &FR + f14Zl + ui

(4)

where subscripts p and n refer to positive and nega- tive rural development areas respectively, B, M, FR and U are percentages of rural municipality popu- lation of British, Mennonite, French and Ukrainian origin. The first two ethnic groups have been gen- erally recognized as successful in terms of con- solidating their economic status within Canada, while the latter two have done less well on average.5 Consequently, the British and Mennonite variables are allocated to the positive development equation, and the French and Ukrainians are earmarked for the negative development equation.

Columns B and D of Table 3 provide the regression results. For those areas of positive development, the machinations of the distance variables are signi- ficantly affected by ethnic factors only in 1961 when the Mennonite variable adopts a direct association with development. By the terminal year, this influ- ence is much diminished. Application of Chow’s test for establishing the similarity of the coefficients implies that there has been a fundamental change in the characteristics of areas of positive rural develop- ment between 1961 and 1971.6 Thus, while distance impacts on the occurrence of positive development have become more pronouced over this period, the influence of ethnic variation has steadily dimi- nished.

On the other hand, calibration of Equation 4 indi- cates that ethnic factors have a larger role to play in the incidence of negative development as the period progresses. Unlike the case for positive develop- ment, Chow’s test suggests that alterations in the characteristics of negative development areas have been more in the nature of changes in degree rather than kind (despite the substantial increase in R* from 1961 to 1971). Hence, these areas tend to be settled by ethnic groups tainted by socio-economic disadvantages. Moreover, the localization of the disadvantaged has tended to become more pro- nounced in pockets of negative rural development as the decade has gone by. In these instances, the negative consequences of separation from smaller centres are being exacerbated by ethnic factors.

Page 9: Observations on the relevance of the industrial-Urban hypothesis for rural development

Geoforum/Volume 14 Number 111983

Conclusion

Analysis of rural development indicators in Man-

itoba has served to demonstrate, in the first place,

that a dichotomy exists between those localities

exposed to metropolitan influences and those sub- ject to influences emanating from smaller urban centres. In essence, metropolitan impacts are not conducive to positive rural development but the impacts of smaller centres are benign, at least in districts of relative proximity to them. In the second place, the results suggest that distance effects have, if anything, increased their hold on rural develop- ment between 1961 and 1971. Finally, the results indicate that factors other than urban accessibility foment rural development. The one other factor explicitly considered, ethnicity, plays an important role in areas of sluggish rural development. These empirical findings call in question the universality of the IUH concept in that they highlight the inappropriateness of accepting the industrial-urban centre as a monolithic whole. Only the smaller cen- tres abide by the positive development assertions of the IUH concept in Manitoba, leaving the large regional centre as an anomaly. At this juncture, a preliminary stage of determining the conditions required for urban growth transmission is sorely in need of integration into the body of IUH thinking. If urban centres are screened initially to ascertain whether or not they have the fabric capable of diffusing spread-effects, the danger of assuming that all cities act as agents of development will be avoided. Unfortunately, as previously noted, growth centre theory is of little help in formulating this initial step even though it lays great stress on the forces responsible for industrial-urban growth.

In contradicting American experience of a link between rural well-being and metropolitan growth, these Manitoba findings have important lessons for rural planning. In a word, they deprecate urban bigness. It is almost an act of faith in growth centre theory to assume that development impulses are geared to the size of urban centre. Yet the Man- itoba results suggest that the smaller centres have more, rather than less, development impacts on rural areas in comparison with the metropolitan centre. By implication, they endorse the unique character of rural development which is specific to time and place and, in so doing, suggest that what is theoretically sound (such as economies of urban size) or empirically demonstrable (such as the effects of metropolitan growth on nonmetropolitan America) is not necessarily apposite to all regions.

This is not to say that regional planners in the Prairies can become complacent about the develop- ment prospects of small centres - the rise in intraregional development disparities between 1961 and 1971 would belie that - but the partial sub- stantiation of the IUH concept augurs well for the promotion of future rural development based on spill-overs from dispersed urban centres of modest scale. However, it must never be overlooked that spill-overs alone are contributing only a minority share to the explanation of the rural development indices (in fact, as Table 3 indicates, between 5 and 24%) and no amount of tinkering to facilitate the promotion of such urban-based stimuli will substi- tute for an appreciation and cultivation of the fac- tors endogenous to rural development. That will only come about when rural development planning is riven from its excessive dependence on the idea of urban centres acting as the panaceas for all develop- ment instigation.

Notes

Perhaps the main exponent of the distrust of urban- centred development in LIPTON (1977). He claims that it imposes a relationship of exploitation in rural areas which cripple their chances for self-sustaining advancement. For a more general critique of develop- ment obsessed with urban dominance, see MORGAN (1978). In some respects, IUH outcomes and those ensuing from Thtinian mechanisms are very similar. For example, high location rents and frequently greater intensity of capitalization on farms close to markets is a standard outcome of Thiinian concepts. However, such results can be construed as a special case of the IUH when the location rent gradient is perceived as a natu- ral corollary to urbanism (i.e. as an urban spill-over). Variously titled ‘spread-effects’ or ‘trickling-down effects’, positive spill-overs serve to transform the rural fabric such that people benefit in terms of improve- ment in incomes, provision of social facilities and gen- eral lifechances. Negative spill-overs (or ‘backwash’ or ‘polarization’ effects) are the exact converse. Under different circumstances, the composite index would have been more accurate if it was the summ- ation of weighted component scores (i.e. allowing for the different strengths of eigenvalues) instead of unweighted ones. However, IUH provides us with no a priori rationale for assigning greater validity to some variables at the expense of others, even though the data-reduction procedure may suggest the contrary. Thus, loadings on all three components are taken at face value to allow for the ramifications of all variables. See, for instance, MAYKOVICH (1975) as well as TODD and BRIERLEY (1977).

6. Chow’s test is construed as an F-ratio test as follows:

F = ( (T-S,-S,)/p)l( (S, + S,)/(n + M - 2~)

Page 10: Observations on the relevance of the industrial-Urban hypothesis for rural development

54 Geoforum/Volume 14 Number 111983

where Tis combined sum of squares (for 1961 and 1971 data), Si is sum of squares attributed to 1961 regres- sion, .Sz to 1971 regression, p is number of variables and n and m are number of rural areas in 1961 and 1971 respectively. The extreme right-hand term denotes row d.f., while p denotes column d.f. in the F distribution. For positive development F = 4.44, significant at 0.01 level; for negative development F = 2.93, significant at 0.05 level but not at 0.01 level. See CHOW (1960).

References

CHOW, G. C. (1960) Tests of Equality Between Sets of Coefficients in Two Linear Regressions, Econometrica 28, 591-605.

CONROY, M. E. (1973) Rejection of Growth Centre Strategy in Latin America Regional Development Planning, Land Economics 49, 371-380.

GILBERT, A. (1975) A Note on the Incidence of Development in the Vicinity of a Growth Centre, Regional Studies 9, 325-333.

GREGOR, H. F. (1979) The Large Farm as Stereotype: a Look at the Pacific Southwest, Economic Geography 55, 71-87.

KATZMAN, M. T. (1974) The von Thuenen Paradigm, the Industrial-Urban Hypothesis, and the Spatial Structure of Agriculture, American Journal of Agri- cultural Economics 56, 683-696.

LAMB, R. (1975) Metropolitan Impacts on Rural Amer- ica, Chicago, University of Chicago, Department of Geography, Chicago.

LIPTON, M. (1977) Why the Poor Stay Poor: Urban Bias

in the World Development, Harvard University Press, Cambridge.

MAYKOVICH, M. K. (1975) Ethnic Variation in Suc- cess Value, In: Socialization and Values in Canadian Society, Vol. 2, pp. 158-179. R. M. Pike and E. Zureik, (Eds). McClelland and Stewart, Toronto.

MORGAN, E. (1978) Falling Apart - The Rise and Decline of Urban Civilization, Abacus, London.

MOSELEY, M. J. (1973) The Impact of Growth Centres in Rural Regions, Regional Studies 7, 57-75.

NICHOLLS, W. H. (1961) Industrialization, Factor Mar- kets, and Agricultural Development, Journal of Political Economy 69, 319-340.

RICHARDSON, H. W. (1976) Growth Pole Spill-overs: the Dynamics of Backwash and Spread, Regional Studies 10, l-9.

ROBINSON, G. and SALIH, K. B. (1971) The Spread of Development Around Kuala Lumpur: a Method- ology for an Exploratory Test of Some Assumptions of the Growth Pole Model, Regional Studies 5, 303-314.

SCHULTZ, T. W. (1953) Economic Organization of Agriculture, McGraw-Hill, New York.

SMIT, B. (1979) Regional employment changes in Cana- dian agriculture, Canadian Geographer 23, l-17.

TAURIAINEN, J. and YOUNG, F. W. (1976) The Impact of Urban-Industrial Development on Agri- cultural Incomes and Productivity in Finland, Land Economics 52, 192-206.

TODD, D. (1979) On Urban Spill-overs and Rural Trans- formation: a Canadian Example, Regional Studies 13, 305-321.

TODD, D. and BRIERLEY, J. S. (1977) Ethnicity and the Rural Economy: Illustrations from Southern Man- itoba, 1961-1971, Canndian Geographer 21, 237-249.