ABSTRACT
Growth pole concepts appear particularly useful to regional planners who are faced with the task of designing a strategy for analyzing and inducing regional growth. This study employs a form of trend surface analysis which identifies urban places that may be interpreted as growth poles and measures trends in growth outward from those poles. An empirical example is provided for the State of São Paulo, Brazil, for the period 1940 to 1960. Two types of growth poles are identified: 1) natural poles that result from a rapid postwar industrialization, and 2) planned poles that result from state governmental efforts to control economic growth.
Key Words:
Notes
1 The authors wish to acknowledge the helpful assistance given by Emilio Casetti.
1 John Friedmann, Regional Development Policy: A Case Study of Venezuela (Cambridge: The MIT Press, 1966).
2 David F. Darwent, “Growth Poles and Growth Centers in Regional Planning—A Review,”Environment and Planning, Vol. 1 (1969), pp. 5 31; and Vida Nichols, “Growth Poles: An Evaluation of Their Propulsive Effects,”Environment and Planning, Vol. 1 (1969), pp. 193 208.
3 Francois Perroux, “Note sur la notion de pôle de croissance,”Economice Applique, Vol. 8 (1955); and idem, “La notion de pôle de croissance,”L'economie du XX an siecle (Paris: Presses Universitaires de France, 1964).
4 Vida Nichols, “Growth Poles: An Investigation of Their Potential as a Tool for Regional Economic Development,”RSRI Discussion Paper, No. 30 (1969), pp. 4–7.
5 Darwent, op. cit., footnote 2, p. 11.
6 J. R. Boudeville, Les Espaces Economiques (Paris: Presses Universitaires de France, 1961).
7 For a detailed discussion see E. Casetti and R. K. Semple, A Method for the Stepwise Separation of Spatial Trends (Ann Arbor, Michigan: MICMAG, 1968). For a computer program, see E. Casetti and R. K. Semple, “GEOFIT,” in Waldo R. Tobler, ed., Selected Computer Programs, Michigan Geographical Publication No. 1 (Ann Arbor: University of Michigan Department of Geography, 1970), pp. 1–18. The program extracts trends from a spatial series by the method described in the example. The input consists of the coordinates of the location of the growth centers and the associated growth. The growth poles and the trends in growth are identified in the computer algorithm in the following way:.
1) A grid outline is established which encompasses all the towns. The outline is partitioned into fifteen equal parts both vertically and horizontally in order to form a 15×15 grid of reference points.
2) From each grid reference point, the distance to every town is calculated and these distances, or some transformation of distances, are correlated with the growth values.
3) A 15×15 matrix of correlation coefficients is formed. The largest coefficient is identified.
4) In order to determine more precisely the actual coordinates of the point of highest correlation, the four diagonal correlation coefficients adjacent to the one identified are used to form the limits for a second finer grid from which a 10×10 matrix of coefficients is calculated as in step 2. Again the largest correlation coefficient is identified.
5) One final 8×8 matrix of coefficients is formed as before. At this stage the position of the largest coefficient is assumed to indicate the actual point of highest correlation on the map. This initial point is identified.
6) The explained variance is calculated. This indicates the variation in urban growth accounted for by the first point.
7) A regression of the growth measures against the distances from the identified trend point to the location of the urban centers is carried out.
8) Residuals from this regression are used as a second set of growth measures.
9) Steps 2 through 8 are repeated, a second point is located, and the trend associated with the point is measured.
10) The procedure continues and additional points are identified until a point and its associated trend account for less than five percent of the variance of the residuals.
8 The two transformations of distance which give best results are the natural logarithm of distance and the reciprocal of distance increased by one. If the first transformation is used the relationship between urban growth and distance from the origin points is of the following form Sij=(ak+bk+ loge dkij) + eij(1).
9 The nine variables were 1) urban population, 2) number of manufacturing establishments, 3) number of employees in manufacturing, 4) value added by manufacturing, 5) number of retail establishments, 6) number of employees in retail activities, 7) total value of retail sales, 8) number of wholesale establishments, and 9) total value of wholesale transactions. Data were specially tabulated by the Conselho Nacional de Estatistica from the Recenseamento Geral do Brazil, 1940, 1950, and 1960.
10 Werner Baer, Industrialization and Economic Development in Brazil, (Homewood, Illinois: Richard Irwin Co., 1965).
11 S. Morley and G. Smith, Import Substitution as an Industrialization Strategy in Brazil, Discussion Paper (Madison: University of Wisconsin Social Systems Research Institute, 1970).
12 Governo de Estado de São Paulo, Plano de Acão, Estado de São Paulo, Imprensa Official do Estado, 1959.
13 Howard Gauthier, “Least Cost Flows in a Capacitated Network,”Geographic Studies of Urban Transportation and Network Analysis, Studies in Geography, No. 16 (Evanston, Illinois: Northwestern University Department of Geography, 1968).