CHAPTER 5 INDUSTRIALIZATION PROCESS AND LEVELS OF REGIONAL-DISPARITIES IN IRAN

Transcription

1 CHAPTER 5 INDUSTRIALIZATION PROCESS AND LEVELS OF REGIONAL-DISPARITIES IN IRAN

2 Ituiustriali:ation.. and ReKiona/IHv)(Jr/ties.. I 09 Having discussed details about the process of industrialisation in Iran and presenting macro perspective as well as sectoral details in this regard, we shall now turn to consider the situation regarding regional disparities in the levels of development and changes in it during the period Our aim is to assess whether the existing gap among regions decreased or increased during recent decades, beginning from 1971/72. For this purpose, quantitative methods shall be given considerable attention in this chapter. 5.1 A Theoretical Review There is, in fact, an important question about the level of regional disparities during the process of national economic development; whether regional disparities converge or diverge with the process of development It may be pointed that there are two sets of forces in this context. On the one hand, expanding economic activities causes spread out of the effects to backward regions. It results in reduction of disparities among the different regions. It may be assumed that the socio-economic condition of the majority people consequently rises to a higher level. On the other hand, although the positive effects of such a development pattern are undeniable, but it simultaneously causes capital and skilled labour to move towards more developed regions because of existence of external economies in those regions. As a consequence, it leads to acute regional divergence. Briefly, the two sets of forces tend towards two different directions from the point of view of regional balance the first tends to reduce inequality and the second increases it. Of course, the final result depends on strength of each of those. Myrdal describes in more detail the forces which he named "spread out effects " and " backwash effects" respectively. According to him, the backwash effects have three components; migration, capital and trade so that the ''localities and regions where economic activities are expanding

3 Industriali:ation... and Regional Di.\fJOrities. 110 will attract net immigration from other parts of the country"'- We should keep in mind that this migration is selective, i.e. included in it are young and skilled labour force, in favour of, rich regions. Similarly, the discussion is true for capital movements that tend to increase inequality because of higher demand for investment in the more developed areas. This in tum will increase incomes, and cause a second round of investment and so on 2. Trade also operates in favour of rich and developed regions against the rest, because the progressive regions are normally industrial and have competitive advantages in comparison to backward areas that remain agricultural. Against the backwash effects there are spread effects moving away from core to the periphery. It means that expansion of economic activities in the more developed areas causes increasing demand for agricultural products, which are produced in backward areas, and therefore, it spurs technical advance there. Myrdal believes that backwash effect is stronger than spread effect in the poor countries, whereas, it is weaker in the developed countries. Consequently, regional inequalities have been diminishing in these relatively developed countries, while the tendency has been towards an increase of disparity in the less developed countries. Moreover, the magnitude of regional inequality in the poor countries is generally much higher than in the richer areas. 3 Hirschman points to "polarization" and "Trickle-Down" effects that are almost the same as backwash and spread effects. Hirschman says there are considerable differences in emphasis and conclusions 4 between Myrdal and himself Generally, it seems that he is more optimistic about the reduction of regional inequalities than 1 Gunnar Myrdal, Economic Development and Under-developed Regions, London; Methuen, 1964,p Ibid. 3 Ibid, p.39. See, A.O. Hirschman. The Strategy of Economic Development, Yale University, 1958, p.187 (Footnote).

4 lndustriali:arion... and Regional /Jispariries. Ill Myrdal. In other words, trickling-down effect is relatively more powerful than polarization effect in the view of Hirschman this bringing about a tendency towards regional equality. Typically, "north region" (Hirschman's idiom for more developed areas) will be specialised in manufactures and the south regions (less developed areas) in primary production.~ In the course of national economic development, the expansion of demand of the north ought to spur southern production. In his opinion, considering low elasticity of demand in the south, the terms of trade move against the There are, at least, three well-known hypotheses regarding the level of regional disparities during the process of national economic development. Some of economists believe that the levels of regional inequalities will rise during the process of development. "Diametrically opposite is the view contained in what may be termed the "Accordion-effect" hypothesis. According to this view, regional disparities converge in the process of national economic development". 7 The third view is named "concentration-cycle" hypothesis; based on this view regional disparities diverge initially and converge in the mature stage Statistical Techniques Used For This Chapter To examine the hypotheses outlined above and other matters of concern of this chapter, various statistical techniques shall be drown upon. Among these, one may refer to the Theil's index, Gini coefficient, coefficient of variation, coefficient of participation rate of labour force, first principal component analysis, and regression analysis. Ibid., p Ibid p.l89. p.3. K.R.G. Nair, Regional Elij)erience in A Developing Economy, Wiley Eastern Limited, 1982,

5 lndustria/i:ation... and Regional Disparities Theil,s Index Theil's index is a useful measure to represent the extent of inequality among regions. It satisfies Pigou-Dalton condition, i.e. a transfer from a rich man to a poor person always reduces the value of index 8. The index can be calculated by using the following formula: Ti =log N- LXi log 1 /xi, where N refers to number of observations ( here number of regions), Xi = Xi!Lxi and Xi is the value of ith observation. Maximum value of Ti equals log Nand its bottom value is zero Gini Coefficient It is attributed to Gini ( 1912), an Italian statistician, and is a well-known inequality measure which has been very widely used to measure income inequality among various groups of people. One way to viewing it is in terms of Lorenz curve ( 1905) in which the percent of population arranged from the poorest to the richest are represented on horizontal axis and the percentages of income enjoyed by the bottom x% of the population is shown on the vertical axis. Obviously 0% of the population enjoys 0% of income and 100% of population enjoy all the income. So a Lorenz curve runs from one comer of a unit square to diametrically opposite comer. Gini coefficient is the ratio of area between the Lorenz curve and the diagonal of the square to the area of triangular underneath the diagonal. If everyone enjoys the same income, the Lorenz curve will be simply the diagonal. So Gini coefficient becomes zero and the coefficient will becomes one in case of perfect inequality. Therefore, normally, Lorenz curve locates under the diagonal with an increasing slope when it moves from a poor section to a rich and richer section 9. There are various ways to calculate Gini coefficient of which one is as follows: 8 A. Sen, On Economic Inequality. C.H. Lewis, Oxford University Press, Delhi, A. Sen. On Economic Inequality. C.H. Lewis, Oxford University Press, Delhi, 1974.

6 lndustria/i:ation... and Re}?ional DivJarities 113 where n stands for number of observations, Yi refers to income enjoyed by each group andy stands for simple average of Yi, whereas y,> Y2 >y,... >yn Coefficient ofvariation(c.v) To measure regional inequalities one of the sensitive techniques for small changes 1s "coefficient of variation". In addition to this, the coefficient passes the Pigou- Dalton condition and is easy to calculate. The formula for weighted (C. Vw) and unweighted (C. Vuw) coefficient of variation are respectively: where Yi stands for per capita income ofthe ith region, Y stands for national per capita income, Un refers to relative population of the ith region, and C. Vuw = [ I(yi-Y) 2 /N] 0 5 /Y where N is number of regions Weighted Coefficient of Participation Rate of Labour Force Analytically, participation rate of labour force or proportional level of employment in each region can be found as one of the major factors, which is relevant in creating inequalities among the regions. Hence, we shall consider the changes of the participation rate of labour at certain points of time. The formula for the weighted coefficient of participation rate is: where rj1 is proportional share of the ith region out of total population, w/pi stands for participation rate of the ith region, Pi refers to total population of the ith region, p = LPi stands for total population of the country and Wi is total labour force of the i,h region Relative Measure: This is a simple measure which presents relative situation of each region in terms,. of per capita income or any selected indicator. It can be calculated by the formula:

7 Industriali:ation... and Regional Dtsparities where Y 1 stands for per capita income of each region and Y N refers to per capita income at national level. If we run the formula, there will be two groups~ the first group includes those regions that have Ri <I 00 whereas the regions of the second one enjoy Ri>lOO. Positive changes of the first group and negative changes related to the second one indicate narrowing-down existing gap among the regions in terms of per capita income during the period under consideration. Moreover, one can calculate the coefficient of correlation between changes of initial Ri and terminal values of it. If it is to be found significant, its sibrn will specify the pattern of changes. 5.3 Administrative Divisions of Iran Often regional studies lead to contradictory conditions since the regions are being taken on the basis of administrative divisions, but for the non-availability of any other data, choice of the divisions is inevitable. The largest divisions of Iran are named "province" and the country has been divided into 25 provinces (Ostan) 10, 229 Shahrestan, 520 Shahr (city) 11. It should be noted that existence of 5,000 population and presence of municipality are the major criteria for identifying a place as a city. 5.4 Regional Divergence In Natural Resources It is worth noting that the provinces differ a lot not only with respect to their area and population but also in terms of enjoying natural resources, rainfall and cultivated area. For example, table 4.1 presents regional disparities in terms of value added of mines among the provinces. 10 Administrative divisions of Iran has undergone some changes during the previous decades. For example, the country had had 22 provinces in 1972 which increased to 23 and 25 provinces in 1976 and 1980s respectively. At present the country includes 28 provinces 11 The ministry of Interior.

8 Industrialization... and ReKional /)i.\paritie.l. 115 IRAN ADMINISTRATIVE DJVISIOHS DY OSTAH :1992 IRAQ Slllboll --~ =~ - _,,,.,...,, '"''",, on.. ~. kgii 0 ~ 100 '" 100,. It- Map 5.a

9 lndustriali;:ation... and Regional Disparities... I 16 Table 5.1 Value added of Operating Mines 1993 Province _\'_!!!~e B:~.d~(~ll _ ra_ils) _.. 0 1~. ()ftotal East Azar West Azar Ardabil Esfahan I 0.6 llam Bushehr Tehran Chaharmohal & Bak Khorasan Khuzestan Zanjan Semnan Sistan & Bal Fars Kordestan Kerman Kermanshah Kohkiluyeh & Boi Gilan Lorestan Mazanderan Markazi Hormozgan Hamedan Yazd Total Source: Data from SCI, SYB, 1373(1994), Calculated by Author. Rank I ll As illustrated by the table, there were large diversities among the provinces in terms of value added of operating mines in Kennan province alone produced 31.0 percent out of total the value added of the mining sector and occupied the first place. On the other hand, Ilam province produced only 0.2 percent out of total the value added and stood at the bottom place from this point of view.

10 Industrialization... and Regional Disparities Value Added of Operating Mines Rla.mn Province I Value Added Fig. 5.1

11 Industrialization... and Regional /Jisparities.. I 18 The existence of natural resources is not always a necessary condition for development of a region and one can find some evidence both intra-country and intercountry to support this assertion. But the important role played by natural resources is cannot be denied. In other words, natural resources create the foundation stone for the process of development of a region. As a result, the prevalence of acute inequality among regions in terms of resource base has potential of creating disparity in the levels of development among them. 5.5 Inequality in Per Capita Expenditure Although state per capita income or State Domestic Product (SDP) may be the most important indicator for measuring regional development, but respective data are not available in case of Iran. In absence of such data, we may select Per Capita Experrditure (PCE) at provincial level instead of provincial per capita income/ provincial domestic product. In fact, per capita expenditure reflects regional disparities in terms of levels of living better than regional per capita income. We have collected data related to per capita expenditure at five points of time, viz; 1972, 1980, 1982, 1984 and 1985 and converted them to 1982 constant prices, using the index of retail prices of the country. It should be noted that selection of the above noted years has been made on the basis of availability of data. Provincial estimates of monthly per capita expenditure are available on the basis of sample surveys conducted by the Statistical Centre of Iran. These estimates are available separately for the rural and urban areas of the country. The figure for each province has been calculated as a simple average of the figures for the rural and the urban areas. Although, weighted average of the figures is more appropriate, particularly in case of provinces in which differences between the share of rural and urban

12 Industrialization... and Re,:;ional Di.\parities I 19 population are important. But in absence of firm data about the percentage of residents in the rural and urban areas at provincial level, we have had to do so. Let us first look at the relative per capita expenditure given in table No.5.2 for each point of time. As mentioned, they are calculated as the per capita expenditure of each province divided by their simple average and multiplied by a hundred. No province reveals a continuous movement in one direction (exception Esfahan). All of them have fluctuated in their direction of movements. If we look at the relative per capita expenditure in the initial period ( 1972) and compare with that of terminal period (1985), the provinces can be classified in two groups. The first group includes those which moved in the downward direction, such as Markazi, Tehran, Gilan, West Azarbiajan etc. The second group contains those which moved in the upward direction, like Bushehr, Chaharmohal & Bakhtiari, Sistan & Baluchestan, Yazd, Ilam and Kordestan. Table 5.2 Per capita Expenditure Relatives Province ~------~--- East Azar WestAzar Esfahan II am Bushahr & Hormoz. Charmohal & Bak. Khorasan Khuzestan Zan jan Semnao Sistan & Balu Fars Kordestan M Kerman Kermanshah Kuhkiluyeh & Boir. Gil an Lorestan Mazanderan Markazi & Theran Hamed an Yazd TotaU Aver Source: The Statistical Centre oflran, Statistical Year Book, Various Years. Calculated by Author

13 Industrialization... and Regional Disparities AVERAGE ANNUAL PER CAPITA EXPENDITURES VA200 LU E (, Rl 100 AL S) PROVINCE Fig. 5.2

14 Industria/i:ation... and Regional Di.\f)(lrities. 121 The weighted and unweighted coefficients of variation have also been calculated for certain points of time. These measures are more appropriate, because they are sensitive to small changes. Table No. 5.3 presents the estimated coefficient of variation of per capita expenditure. Table 5.3 Coefficient of Variation of Per Capita Expenditure Year ~-~ig~te_djc:~ YJ !J_~~~igh ~ed (C. V) The table reveals that despite of fluctuations in the weighted coefficient of variation, the trend shows decline from the year 1972 to 1982 although again th~re is an increase from 1984 to But over the thirteen-year period as a whole these appear to be some decline in regional per-capita consumption disparities. Gini coefficient is another measure to represent the extent of inequality among the regions. The estimated values of the coefficient for the five points of time are Table 5.4 Gini Co-efficient of Per Capita Expenditure Year Gini Coefficient ~ As shown by the above table, the Gini coefficients indicate an increase from to during the period It shows a substantial reduction in 1982 but the coefficient went up to after two years. If one compares two terminal points of

15 Industrialization... and Regional Di.v}(Jrities 122 time, 1972 & 1985, inequality in per capita expenditure decreased slightly among the provinces in 1985 in comparison to that in 1972 in terms of Gini coefficient. These figures do not indicate any clear trend of regional disparities; however, it is worth noting that the coefficients have somewhat small values at all points. It gives some indication that interprovincial inequalities, irrespective of its trends, have not been very serious in terms of the selected indicator over the period under consideration. Theil's indices reveal even smaller values than those of the co-efficient of variation and Gini coefficient. The high value of Theil's index is only 0.01 pertaining to the year of 1972 whereas the smallest value is in case of Table No. 5.5 compares the values and trends of the coefficient of variation, Gini coefficient and Theil's index at the specified points of time. Table S.S Inequality Measure Year Weighted Unweighted(C Gini... _(~~Yl...Yl ~~~!fi~i_e_f!~ Theil's Index As the table reveals, all the above criteria indicate that inter-provincial inequalities decreased in terms of per capita expenditure in the terminal year( 1985) in comparison to the initial year ( 1972) though only be a small magnitude.

16 Industrialization... and Regional Disparities _..,_ Wcv --unwcv b. Gini Co -4 Theil's lnd YEAR Fig. 5.3 Regional Inequality in Per Capita Value-added of Manufacturing Activities To take into account regional disparities, in absence of data pertaining to per capita income at provincial level, we have already examined position regarding percapita expenditure. We now turn to per capita value-added in manufacturing industry. This indicator is important because it can be taken to reflect the level of industrialisation and so examining the trend of this indicator can gouge a major part of the effects of industrialisation upon the inter-provincial divergence. For this purpose, we have estimated Gini coefficient, Theil's index and coefficient of variation of this indicator at certain points of time namely, 1972, 1976, 1986, 199land The results have been shown in table No. 5.3 as follows:

17 Industrialization... and Regional Disparities Table 5.6 Inequality Measures For Per Capita Value-added Of Manufacturing Activities Year c.v Gini Co Theil,s Ind The Coefficients of 1993 have been calculated based on Per Worker Value-added In That Year. First, it may be noted that the magnitude of regional disparities in terms of this indicator of industrialization is far higher than that in terms of per-capita expenditure. Further, as the table illustrates, inter-provincial disparities have continuously decreased in terms of both coefficient of variation and Gini coefficient. However, Theil,s index reveals an increase during the period In any case, all the three measures indicate that inequality in terms of these indicators has drastically decreased among the provinces in 1993 compared with1972. Trends of Inequality Muauret~ (In t.nna of manufac. pcva).. f 1. 2 r-~~~~~~--~~--~--~~~--~~~~~--~~~~~ :!..! 1 r-~~~~~~----~r-----~~~~~~----~~~--~ ~ o.s r-~--~~~~~ ~~--~~~~~--~~~~~~~ (J 0.8 ~.j... ~-~~~==::~~==~~::==~~~==~==~~------~~~~ Fig c.v - Gini eo. Theil'~

18 Jndusfriali::ation... and Regional Disparities Responsible Factors Related to Regional Disparities As mentioned earlier, participation rate of labour force i.e. the proportional of employed population of each region to its population can be taken as an important factor, which can account for a major part of disparities among the regions. So the weighted coefficient of participation rate of labour force has been calculated to look at the tendency of provincial divergence form this point of view. Table 5.4 presents the participation rates at four given points of time. Table 5.7 Participation Rates of Labour Force J>rovl~-- _ - -!l~(!~i!l. --!~~{1_9_76)_.. 11_6~{!~L- _ _371!_(1~_1) East Azar West Azar Esfahan Ilam M Bushahr & Honnoz Channohal & Bak Khorasan 2! Khuzeotan 18.! ! Zanjan Semnan 23.!1 2! ! Sbtan & BaliL ! Fan ! Kordestan Kennan ! Kennanshah !1 Kuhklluyeh & Bolr Gilan Lorewtan Mazanderan Markazl & Theran !1.7 Hamedan ! Y azd ! !1 _'!.!.tall ~!~~ ~2.9.. z_~l- 2_;~...!~ s_ c.v "'' Source: Data From SYB oflran, Various Years Calculated By Author. The coefficients of variation of participation rate have been given at the bottom of the table. As may be seen, although the coefficient shows fluctuations during the period under consideration. It decreased from 0.20 in 1972 to 0.11 in 1976 but again it increased to 0.27 in 1986 and finally it reduced to 0.11 in Hence, the coefficient indicates some decrease in inequality among the provinces in 1991 in comparison with The period covered is, however, too short to make as assessment of long run trend.

19 Industrialization... and Regional lji!>parities. 126 In order to examine the effects of industrialisation process and the level of employment on regional disparities, an attempt has been made to regress 12 per capita expenditure (as a representative of per capita income) on participation rate of labour force and per capita value added of large scale manufacturing industry sector at different points of time. It should be noted that the regression equation has been estimated in two stages, with one-year lag. At the first stage, per capita expenditure (at constant prices) in 1972 (PCE72) has been selected as dependent variable and participation rate of labour force in 1971 (PRL 71) and the per capita value added of large scale manufacturing establishments (PPW1) have been given as independent variables. The estimated equation is: PCE(72)= PPWI(71) PRL(71) (4.85) (2.86) (0.54) Multiple R = 0.56 R 2 = 0.32 D.W = 1.86 Figures in parentheses are ltl values. Significant at less than 1% level Whereas the independent variable of PPWI(71) and constant tenn are significant at 1% level of probability, the other independent variable PRL(71) is significant at a low level (59% ) of probability. In addition to this, the variable has an unexpected negative sign in the estimated equation. It can be said that the independent variables are free from auto-correlation problem on account ofd.w = 1.86, but estimated R 2 = 0.32 indicates that explanatory power of independent variables are not strengthen sufficiently. At the second stage, we have added the variable "average wage and salary" in the large scale manufacturing establishments in 1972 ( A W72) to the given equation 12 Cross-sectional regression has been applied in absence of sufficient data to run time series regression.

20 Industrialization... and Regional Disparities as an independent variable. The estimated equation including the new explanatory variable is as follows: PCE(72) = AW(71) PPWI(71) PRL (3.89) (1.59) (2.05).. (0.27) Multiple R = 0.64 R 2 = 0.41 D.W = 1.95 Figures in parentheses are ltl values. Significant at less than I% level. Significant at less than 5% level. The latter equation reveals that the added explanatory variable (A W72) is an effective variable. It has increased the explanatory power of the estimated equation from 32% to 41 percent. Moreover, the estimated coefficient of the variable has an expected positive sign with a low standard deviation (0.03). The independent variables of PPWI(71), AW(72) and constant term are significant at 5%, 13%, and 1% respectively. However, PRL(71) again has a negative sign and is significant at a very low level (79%) of probability. The above estimates show that the variables of AW(72) and PPWI(71) can account for an important part of variation of the dependent variable (PCE72). Moreover, they have been taken from manufacturing industry sector. It can therefore be inferred that there is a significant relationship between variation in per capita expenditure and that of industrial activities. In other words, one can perhaps attribute to industrial activities the role of a factor which has considerable potential of creating regional disparities. Alternative equations have also been estimated, in the above manner with lag and without time lag, to inquire into relationship among the specified variables in the years 1980, 1982 and Results obtained by running cross-sectional regressions revealed that no significant relationship can be found between dependent variable PCEso on the one side and independent variables PPWlso and AWso on the other.

21 !ndustriali:ation. and Regional Di.\parities When logarithmic and lagged forms of the equations were regressed, no improvement was found in the explanatory power of the new regressions. PCE(80) = PPWI(80) A W(80) (5.3) (-0.03) (0.89) Multiple R = 0.25 R 2 = 0.06 D.W = 2.18 Figures in parentheses are itl values. Significant at less than I% level. As may be observed from the above equation, apart from constant term, which is perfectly significant, the other coefficients have a very low level of confidence. Coefficient of determination also accounts for a very low percentage of variation of the dependent variable and small value of multiple R indicates a weak correlation between dependent and independent variables. It is time now to look at the results obtained m the light of application of regression technique, with one year lag, for the year Estimated coefficients have the expected positive signs and R 2 =0.19 is more powerful than that for Despite this, " t " values are relatively small and so the estimated coefficients, with the exception of the constant term, enjoy low levels of confidence. A summary of estimated coefficients is presented below: PCE(82) = A W(81) PPWI(81) (5.3) (1.4) (1.3) R 2 =19% D.W=1.66 Multiple R = 0.44 Finally, a lagged model has been used to consider the relationship for the year The estimated equation is given below: PCE(84) = A W(83) PPWI(83) (2.1) (1.1) (1.3) R 2 = 20% D.W = 1.82 Figures in parentheses are ltl values. Significant at less than 1% level. usignificant at less than 5% level

22 Industrialization... and Regional Disparities Although the estimated coefficients have positive 'signs and "R 2 " can explain 20% of variation of dependent variables but "t" values are small and hence the coefficients are significant at low levels of probability. Table 5.5 contains of a summary of the estimated equations in the mentioned years. Table 5.8 The Estimated Equations of Per Capita Expenditure Equation R2 Sig. T% D.W PCE(72)= PPWI(71) PRL(71) 32% 1, I, PCE(72) = AW(71)+ O.ot PPWI(71) PRL(71) 41% 1, 5, PCE(80) = PPWI(80) AW(80) 6% 0, 91, Log PCE(82) = Log AW(82) Log PPWI(82) 2% 3, 55, PCE(82) = AW(81) PPWI(81) 19% 0.00, 18, PCE(84) = AW(83) PPWI(83) 20% 5,30, As illustrated by the above table, one can say that industrial variables can account for an important part of provincial disparities in terms of the selected variable (PCE) during the period under consideration. In fact, the case of 1980 is a result of severe reduction of industrial activities in the country in that period on account of occurrence of upheavals of the Islamic Revolution in early 1979 and out-break of Iraq-Iran War in Regional Structural Changes during Industrialisation Process in Iran It is generally known that one of the most important effects of industrialisation process on a given economy is emergence of structural changes in that economy. The changes may take place in production system, production composition, structure of exports and the share of each sector out of total employment. This matter has received considerable attention of well-known economists, so much so that structural changes has often been taken as the most important criterion by which an economy can be

23 lndustriali:ation.. and Re!(iona/ Di.\fXlrities called an industrialised one. For instance, the Clark-Kuznets hypothesis points out that lower share of agricultural sector in employment or in national income would indicate a higher level of development of that economy and vice-versa. Hence, it should be borne in mind that although per capita income or state per capita income has received the greatest attention for evaluating the level of development or the stage of industrialization in which a country is placed, but structural indicators are equally important from this point of view. This is all the more important, because there are a number of countries which have high per capita income, as high as almost that of developed countries, but these countries are not taken as industrialised societies. The matter of structural changes has already been considered attention in the previous chapter from the point of view of production composition as well as from the angle of structure of exports during the industrialisation process of the country. An attempt is now made to study this issue from the viewpoint of sectoral distribution of labour force at provincial level. Since the percentage of labour force engaged in industrial activities has been taken as an important indicator of industrialisation in the literature by most of the scholars, the proportion of industrial labour force in the provinces can be taken as a measure of the provincial level of industrialisation. Therefore, we shall consider the matter of structural changes of the country from this point of view. Table No. 6 contains the provincial percentages of employment in the primary sector at four points of time namely, 1971, 1976, 1986 and 1991.

24 Industrialization... and ReKionai!Jivwrities. 13\ Table 5.9 Percentages of Employment in Primary Sector. Pr~yi_ll~~ East Azar Rank ll West Azar Rank Esfahan Rank Dam Rank Bushehr& Hor g 21.7 Rank Charmohal& Bakh. Rank ll 13 Khorasan Rank l2 Khhzestan Rank Zanjan Rank Semnan Rank l Sistan& Balu Rank Fars Rank Kordestan Rank Kerman Rank Kermanshah Rank Kubkiluyeh&Boi Rank Gilan Rank Lorestan Rank Maznderan Rank Markazi& Tehran Rank Hamed an Rank Yazd Rank Total/Ave Source: SYB, SCI, 1351(1972), 1360(1981). 1370(1991) As shown by the table, the share of primary sector m employment has drastically reduced during the period Based on Kuznet's thesis, the

25 Industrialization... and Regional /Ji.v)(lrities country has therefore moved along the path of economic development. The sector had the lowest share out of total employed workforce in the five provinces; Markazi & Tehran, Yazd, Esfahan, Khuzestan and Semnan. Hence, these five can be placed in the highest level of development in terms of employment structure. The relative position of provinces shows remarkable changes in many cases (except six provinces which kept their ranks). For example, the ranks of Ilam, Hamedan and Kerman have changed amounting to a shift of 12, 7 and 7 places respectively in the course of the period Rank correlation was estimated between 1971 and 1991 at three different intervals of time as given below: Rank correlation between 1971 and 1976 = 0.78* " " " 1971 and 1986 = 0.89* " " " 1971 and 1991 =0.78* 13 The study shows, all rank correlation coefficients are positive and high. A comment on the implication of this correlation needs to be made. The correlation coefficient between 1971 and 1986 is high and indicates that provincial disparities recorded a slight reduction in terms ofthe proportion of workforce engaged in primary sector. The coefficient of variation of percentages of employment in primary sector were estimated for four points of time and these are presented as below: Year c.v The coefficient of variation initially increased from 0.40 to 0.46 in 1976, and then it has decreased to 0.39 during The coefficient of variation shows a slight decline in the terminal period compared with the initial period. It can be said that interprovincial disparities have shown some tendency to decline during the period under study but not a very pronounced one. Spearman's Rank Correlation is calculated by the formula: r.= (6l(A-B) 2, where A and Bare ranks of variables in initial and terminal period, n is number of observations (here no. of provinces). 13 * Significant at I% level.

26 lndustriali:ation.. and Regional Disparities. 133 An attempt is now made to examine the magnitude of structural shift from the point of view of the share of primary sector in the workforce. Structural change has been measured as percentage of workforce engaged in primary sector at a point of time minus that at the next point of time, i.e., Pt - Pt+ 1, where Pt is percentage of labour engaged in primary sector in a particular census year and Pt+ 1 is percentage in the next census. Its main advantage is that it yields a positive sign in case the proportion of primary sector workforce declines, which reflects a progressive shift of the sectoral structure (Ashok Mathur, 1994). Table number (6) shows change in the primary sector workforce percentage. A quick look at it reveals important structural change, particularly in the period The last column of this table indicates the composite structural change over the period As shown there, Ilam has had the highest percentage of structural change, Kuhkiluyeh and Boirahmad, Kordestan and Zanjan followed it respectively. It is noticeable that maximum structural change occurred during the period As a consequence, the percentage of workforce engaged in primary sector declined from 46.7 per cent (in 1971) to 34.0 per cent (in 1976) at the country level. However, the percentages of structural change were 5.0 and 4.5 during the next two periods respectively. The proportion of labour engaged in the non-agricultural sector has increased almost continuously across all the provinces during the period , although the both magnitude and direction for industry differ among the provinces.

27 Industrialization.. and Regional Disparities. 134 TABLE 5.10 CHANGE IN PRIMARY SECTOR WORKFORCE PERCENTAGE Province Rank Rank East Azar We5t Azar Esfaham II am Bushehr & Hor to Chaharmoha1 & Ba Khorasan Khuzestan Zan jan 25.7 I Semnan Sistan & Ba1u Fars Kordestan Kerman Kermansbah Kuhkilu. & Boi Gil an Lorestan Nazanderan Markazi & Teh Hamedan Yazd ' TotaUAve Province Rank Rank East Azar West Azar Esfaham Ham I Bushehr & Hor Chaharmoha1 & Ba Khorasan Khuzestan Zan jan Semnan Sistan & Balu Fars Kordestan Kerman Kermanshah Kuhkilu. & Boi Gil an Lorestan Mazanderan Markazi & Teh Hamed an Yazd TotaUAve The share of industry and non-agricultural sectors are given in table No As shown there, Y azd had the highest rank from the point of view of proportion of workforce employed in industry and second rank in non-agricultural workforce over

28 !ndustria/i:ation... and He~ional Disparities the period under consideration, followed by Esfahan, Markazi & Tehran and Khorasan in terms of proportion of workforce employed in industrial sector. West Azarbaijan, Gilan, Kuhkiluyeh & Boirahmad and Ilam occupied the lowest ranks from this point of view. Table 5.11 SECTORALSTRUCTURE:PERCENTAGEOFEMPLOYMENTIN INDUS. & NON-AGRI. SECTORS Province Indus. Non-Agri. Indus. Nong-Agri East Azar Rank II 13 5 West Azar. II RJutk Eafaham Rank flam RJutk Bwbehr & Hor Rank Chaharmohal & Ba Rank Khonuan Rank Khuzestan Rank Zanjan Rank Seaman Rank Slatan & BahL Fan Rank IS Rank Konlestan Rank Kerman Rank Kermanshah !1.9 Rank 14 It IS Kuhkilu. & 8oL Rank Gllan Rank Lorettan Rank Nazanderan Rank M.vkazl & T eh Rank 4 I 9 Hamed lui ! Yazd RJutk Rank 1 2 I Total/Ave ~.J tl 48.4 t IS I Contd..

29 Industrialization... and Regional Di.IJXlrities. 136 Province 1986 Indus. Non-Agri. Indus. East Azar Rank West Azar Rank Esfaham Rank Ham Rank Bushehr & Hor Rank Chaharmohal & Ba Rank Khorasan Rank Khuzestan Rank Zanjan Rank Semnan Rank Sistan & Balu Rank Fars Rank Kordestan Rank Kerman Rank 10 II Kerman shah Rank Kuhkilu. & Boi Rank Gilan Rank Lorestan Rank Nazanderan Rank Markazi & Teh Rank 6 I 3 I Hamedan Rank Yazd Rank TotaUAve

30 lndtl.\'lrialization.. and Regional Di.\paritie.L 13 7 Rank correlation coefficients between , and for industry and non-agricultural sectors are presented below: Rank correlation between 1971 and 1976 Rank correlation between 1971 and 1986 Rank correlation between 1971 and 1991 Indus Non-agri The figure shows that all rank correlation coefficients are positive and relatively high. Therefore, it can be said that broadly a similar hierarchy of interprovincial disparities has continued over the period. As already mentioned, coefficient of variation is more sensitive to small changes. So in order to examine the pattern of changes of provincial disparities, the estimated coefficients of variation for percentage of workforce engaged in the industry and tertiary sectors are given respectively as follows: "Coefficients of variation of percentage of workforce engaged in industry and tertiary sectors" Year Industry Tertiary On the basis of this table one can say that interprovincial disparities have decreased in both industry and tertiary sectors. It may be noticed that a reduction in interprovincial disparities has been continuous and more pronounced in the tertiary sector than in the industrial sector, because it decreased from 0.42 in 1971 to 0.14 in 1991 in tertiary sector as compared with 0.44 to 0.26 in industrial sector during the same period. 5.9 Principal Component Analysis As pointed out in discussion of methodology used in the thesis, one of the statistical techniques, which will be used to examine provincial disparities, is the method of Principal Component analysis. The method of first Principal Component is

31 Indu.wrializarion.and Regional /Ji.IJKlrifies 138 a useful method for obtaining a comprehensive view of the levels of development among the provinces. This method enables us to construct a composite index and to compare the provinces in terms of given variables through this new index. The reason why we have used composite index is the fact that it may be a province shows more advantage from the point of view of one indicator, say percentage of urbanization than others but it holds a low rank in terms of another indicator. Therefore, selecting one indicator cannot give a very true picture of overall disparities. The composite index obtained by this method is characterized as having the maximum sum of squares of correlation of the index with the given variables. On account of this property, it may be considered as the best representative of all the given variables'" 4 The method is not suitable in case of low correlation among the variables. In order to construct composite index, using principal component, after computing the highest eigenvalue, we must compute the corresponding eigenvector, and then multiply the elements of the eigenvector into the standardized values ofvariables. Keeping in view the limitation of available data at provincial level, two different ways have been selected to construct composite indices by using the method as follows: 1. We have chosen I 2 variables including industry, agriculture, infrastructure and urbanisation and ran the method of first principal component analysis at four points of time viz, 1972, 1976, 1986 and 199 I. After computing composite indices of the provinces, we have categorized them in four classifications. 2. To apply principal component analysis in order to construct composite indices, the economy of the country has been viewed in terms of the agricultural sector and non-agricultural sectors (which includes industry, tertiary, infrastructural and urbanization sectors). We have selected 2 I variables at two points of time viz, 1972 and After calculating composite indices for the agricultural and non- 14 Aslam Mahmood; Statistical Methods in Geographical Studies, Rajesh Publications. p.99, 1988.

32 Industrialization and Re~;ionaf 1-'i.~pantie.\ 139 agricultural sectors, we have combined them in a weighted linear combination manner Construction of Composite Indices by Usine Pooled Correlation Matrix As pointed out, in order to calculate composite indices by using first principal component analysis in this manner, twelve variables from different sectors have been taken as follows: PPA =productivity per acre in major wheat production in selected years, FC= number of factories per one million population "' = = PCV =per capita value added in large scale manufacturing establishments, PWV = the value added of per worker in the large scale manufacturing units, A W = average wage and salaries of labour force engaged in the manufacturing units, BED= number of beds per 100,000 population, DOC = number of doctors per 100,000 population, ST = number of secondary students per 100,000 population, VS = number of vocational/ technical students per 100,000 population, PA =number of postal agencies = = = ER = the length of existing roads per one square kilometre, UP = percentage of urban population. Results obtained by applying first principal component methodology in case of 1972 data are given below: Correlation Matrix: AW72 BED73 DOC73 ER7J FC7J PA72 PCV72 PF7J PPW72 STU73 UP66 VS73 A W72 1,00000 BED73, ,00000 DOC73,2469~, ,00000 ER73, , , ,00000 FC73,17689,71471,750~ -, ,00000 PA72 -,23247,34903, ,36~98, ,00000 PCV72,32039,82126, ,11241,66681, ,00000 PF73 -,32908,1)8450 -,0406~ -,!5212~,14566, , ,00000 PPW72,37983,74207,81788,02~~1,!58401,10777,80~98 -,29322 STU73,34374,81663, ,134~,67343,22733,6~298,0667~ UP66,27090,89203, ,38090,75893,473~1,76030,16017 VS73,060~7,70237,678~7 -,44192, ~487,!57338,25187 Source: Data from SYB, Various Years, Calculated by Author. 1,00000, ,00000,749~9.7~030 1,0000,41072,6~49.72~9 1,000 As shown in the correlation matrix, three variables, namely, postal facilities (PF73), productivity per acre in major wheat(pa72) and length of roads(er73), have negative correlation with some other variables. The others have positive and relatively high correlation. 1 ~ The method has been used by M.N Pal in order to study regional disparities in case of India in 1975.

33 !ndustria/i:ation... and Regional!Ji.\parities 140 The highest eigenvalue is 6.50 which explains 54.2 percent of total variation and the corresponding factor matrix (factor 1) also is given: Factor Matrix: variable Factor 1 UP66,939ll BED73,93092 DOC73,92917 PCV72,87337 fc73,85438 ST73,84021 VS73,81214 PPW72,807l9- PF73 -, ER73 -,31198 PA72,44027 AW72,28729 Signi~cant ~t 1% probabi~tt The fonnula which has been used to test th~ significance of corr.elation coefficients 1s: t = r(n-2) /( 1-~) 05 where n = the No. of observations, r = the coefficient of correlation and n-2 stands for degrees of freedom. The factor loadings of first principal component show that it has a significant positive correlation with eight variables; percentage of urbanization, No. of doctors, No. of beds, per capita value added of manufacturing establishments, per worker value added of manufacturing units, No. of factories, No. of secondary students and No. of vocational/technical students. It also has a negative significant correlation with postal facilities. This first principal component may be called the index of " provincial development" in Iran. According to results obtained by runmng the method of first principal component analysis, we shall arrange the provinces in four classifications; developed group. relatively developed group, less developed group, and underdeveloped one. The value of composite indices will be used to classify the provinces. Those province~ which have composite indices greater than one shall be classified in developed group, provinces with scores between one and zero (l>score>o) in relatively medium developed group, provinces with scores smaller than zero and greater than minus one (O>score>-1) in less developed one and scores less than minus (score<-1) one will be

34 lndustna/ization... and Regiona/J)i.,f}(1rities. 141 placed in underdeveloped group. Table No presents classification of the provinces in the above groups as follows: Table 5.12 Classification of Provinces 1972 ~Prov_i~_ce_ Composite Index (score). Ranks A: /2eJ elof!.ed ~rouf!. Tehran & Markazi Khuzestan B: Relativelr_ Develo11.ed (ir:.ouf!. Esfahan Yazd Kordestan Mazanderan Fars Gilan Semnan Khorasan C: Less ~elof!.ed GrOf!.f!. West Azarbaijan ll Hamedan East Azarbaijan Kermanshah Kerman Lorestan Bushehr Chaharmohal & Bakhtiyari Kokiluyeh & Boirahmad Sistan & Baluchestan Zan jan D : Cf..nderdevelo~ fi.r.ouf!. Hormozgan Ilam Let us now look at the results obtained due applying to the first principal component analysis for 1976: Correlation Matrix (1976) AW76 BED76 DOC76 ER76 FCS76 PCV76 PF76 PA76 PPW76 ST76 UP76 VS76 AW76 BED76 DOC76 ER76 FCS76 PCV76 PF76 PA76 PPW76 ST76 UP76 VS76 1,0000 1,00000., ,00000,37919, , ,00503,28629, ,00000,20624,92100,79021, ,00000,40312,83456,92760,28721, , ,21969,07674, ,35120,04538, , ,06315,04723, ,22048,01961,00609, ,00000, ,34537,01381,15855, , , ,00000,40421,75639,59994,23574,59428, , ,07299, ,00000,41799,89881,81301,12046,83586,73976,09156,13624,32385, ,00000,21291,64447, ,01906,58340,54784, ,03587,61017,53582

35 Industrialization... and Regional /Ji.1parities As the correlation matrix illustrates, the variables of length of roads, postal facilities and productivity per acre in major wheat have insignificant negative correlation with some other variables. The biggest eigenvalue is which explains 48.0 percent of total variation, whereas it was founded to be as high as 54.2 percent in the The corresponding factor matrix is given as follows: Factor Matrix: Factor 1 AW76,49324 BED76,96338 DOC76,92982 ER76,26581 FC76,86784 PCV76,90229 PF76,04607 PA76,04384 PPW76,42940 ST76,78709 UP76,89337 VS76,69~87 The factor loadings of first principal component have positive significant correlation with average wages of industrial workers, No. of beds, No. of doctors, No. of factories, per capita value added of manufacturing firrns(large scale), postal facilities, productivity per acre of major wheat, productivity per worker of large scale manufacturing establishments, percentage of urban population No. of vocational/technical students and No. of secondary students. The provinces have been arranged into four categories by using their composite indices as follows:

36 Industrialization... and Regional /Jisparities _l~t:c!vinc~ A: Developed Group Tehran & Marlulzi Semnan Yazd B: Relatively Developed Group Esfahan Khuzestan Gil an Fars Kerman Khorasan Mazanderan C: Less Developed Group Kermanshah West Azarbaijan East Azarbaijan Bushehr Chaharmohal & Bakhtiary Lorestan Hormozgan Sistan & Baluchestan Hamedan llam D: Underdeveloped Group Kordestan Kohkiluyeh & Boirahmad Zan jan Table 5.13 Classification of provinces ( 1976) Composite Inde~(score) 3.39 I Ranks Comparison of the provincial scores in the above table and that of the table No. 5.8 related to 1972 reveals an upward movement of Semnan and Y azd provinces from the relatively developed group to developed one, whereas Khuzestan moved in the reverse direction. Meanwhile, Ilam and Hormozgan improved their ranks by one level and moved into the less developed group from underdeveloped one. Kohkiluyeh & Boirahmad, and Zanjan provinces descended to the underdeveloped one in Semnan went up to the rank of 2 in 1976 from 8 in Therefore, it can be said that the level of living deteriorated in case of Kordestan and it improved pertaining to Semnan and Yazd in Trends of Composite Indices during the Post-Revolution Phase As pointed out, the method of first principal component analysis will be used to calculate composite indices at two points of time, 1986 and 1991, in the postrevolution phase. It should be noted that selection of the above periods has been done

37 Industrialization... and Regional /Ji;,parities in the light of available data. The results obtained by using a pooled correlation matrix are given as follows : Correlation Matrix (1986): AW86 BED86 DOC86 ER86 FC86 PCV86 PF86 PA86 PPW86 ST86 UP86 VS86 AW86 BED86 DOC86 ER86 FC86 PCV86 PF86 PA86 PPW86 ST86 UP86 VS , ,00000,47742,81~98 1,00000, , , ,00000,37062,69269 ~9~8 -, ,00000,62891,61807,6!1754,11303, , , , , ,..W392, , ,00000,34343,44320, ,35161,70501,46796, ,00000 ~84!15, ~... -,06261 ~ll87, ,16667 ~8903 1,00000,44519,39707,50027,36248,30526, ,05510,15451, ,00000,41910,'19153, ,13094,68227, ,07418,45891,45184, ,00000,11199,66659, ,12064,5!1187,39440,1822,17739,34384,64314, ,00000 The highest eigenvalue is 5.6 which explains 47.1 percent of total variation. The corresponding factor matrix is: Factor Matrix: BED86 OOC86 PCV86 FC86 UP86 PPW86 ST86 PA86 AW86 ST86 ER86 PF86 Factor,87514,83548,83346,83~,82174,75654,64604,64213,61805, , ,10659 Significant at 1% level. Significant at 5% level. It should be noted that with the exception of the variables of length of roads and postal facilities, the other elements of factor matrix are significant at a high level. On the other hand, higher value of variables in the factor matrix would correspond to the higher level of development, because they are taken as weights in constructing composite index. The factor loadings of first principal component analysis show that it has a significant correlation with ten variables. We therefore call it as the index of provincial development in 1986.

38 lndustriali::ation... and ReKional Di.\parities. 145 As mentioned before, we have classified the provinces into the four categories on the basis of estimated composite indices as follows: Table 5.14 Classification of Provinces (1986) _fr~~lnce A : Developed Group Tehran & Markazi Yazd Esfahan B Relativelv Developed Group Semnan Fars Kerman Zanjan Khuzestan C :Less Deyelqped Grouo Mazanderan Khorasan East Azarbaijan Gil an Chaharmohal & Bakh. Kermanshah Hamedan Lorestan Bushehr West Azarbaijan Lorestan Kohkiluyeh& Boir. D: Underdevelooed Group Kordestan Ham Sistan & Baluch R1lnk A comparison of the ranks of provinces reveals some changes from the point of view of rank ordering when we move from 1976 to 1986 namely, Khuzestan lost its rank from 5th in 1976 to 8th in 1986 followed by Sistan & Baluchestan who lost by four places. On the other hand, Tehran & Markazi and Semnan provinces remained in the top category, Zanjan improved drastically from this point of view. It improved its rank from 23 (the bottom place) in 1976 to in Let us now turn to study interprovincial disparities on the basis of first principal component analysis in case of 1991.

39 Industrialization. and Regional /)i.\fjaritie.\ Correlation Matrix (1991) AW91 BED91 DOCJI ERS'JI FC91 PCV91 PF91 PA91 PPW91 ST93 UP91 VS93 AW91 1,00000 BED91 -, ,00000 DOC91,1~951,6~87 1,00000 ER91 -, ,03263,0533~ 1,00000 FC91 -,06543,56~15,62668,28~ 1,00000 PCV91,02..~,43390,39834,04495,67~1 1,00000 PF91,61070,08740, ,02111, , ,00000 PPA91 -,12944,4~19,37278,06903,61108 ~725, PPW91,69100,06298, , ,07796,29000,44780, STU93 -,140~1,19880,21867,2794~,19302,05905,04823, , UP91 -,20386,62410,58586,oom,.'i8~14, ,00313,46932,01785, VS93 -,49793,.'19194,41902,20700,.'13864,2~82 -,06114, ,3588~,46730, The highest eigenvalue is 4.2 which explains only 35.3 percent of total variation whereas the percentage was 54.2 percent pertaining to It gives some indication that the explanatory power of selected variables was stronger in 1972 than inl991. The corresponding factor matrix is given as follows: Factor Matrix: FC91 BED91 UP91 DOC91 PA91 VS93 PCV91 ST93 AW91 PPW91 PF91 ER91 Fgctor 1,86783,79441,77046,74902,70072,7006~, , , ,04368,11490,20527 Significant at I% level. The factor loadings of the first principal component analysis show that it has significant positive correlation with seven variables. It may be called the index of provincial development in We have arranged the provinces into the categories as given in table 5.10.

40 lndustria/i:ation... and ReKional /Ji.~parities Table 5.15 Classification of Provinces ( 1991) _Province... A; Developed Group Tehran & Markazl Yazd Esfahan Semnan 8: Relatively De eloped Group Kennan ZanJan Fan Buabehr C; Less Developed Group Chaharmohal & Bakh. KhUUtltan Khoruan Muanderan Kennanshah Gtlan East Azarbaljao Honnozgan llam Lorestan Hamed an Kuhklluyeh & Boir. D; Underdeveloped Group Kordestan West Azarbaijan Sistao & Baluch. <;:omposuc lndu(!korc) ! ! ! !12-0.! Rank I 2.J 4!I 6 7 H 9 10 II ! A look through the value of composite indices of the provinces in 1986 and 1991 reveals some changes in placements among the provinces. Meanwhile, Tehran & Markazi, Yazd and Esfahan remained at the top. A few provinces such as Bushehr, Ilam and Chaharmohal & Bakh. improved their ranks by as many as 9, 5 and 4 orders respectively in 1991 compared with their positions in On the contrary, some other provinces like West and East Azarbaijan, Mazanderan and Khuzestan lost their places by as many as 4, 4, 3 and 2 places respectively in 1991 in comparison to 1986.

41 Industrialization... and Regional Disparities Levels of Provincial Development Province Fig. 5.5 lcc.l1972 c.l1976 Dc.l1986 oc.l1991 l 1. East Azarbaijan 2. West Azarbaijan 3. Esfahan 4.11am 5. Busbebr 6. Chaharmobal & Bak. 7. Khorasan 8. Khu.zestan 9. Zanjan 10. Semnan 11. Sistan & Balu. 12. Fars 13. Kordestan 14. Kerman 15. Kermansbab 16. Kubkiluyeb 17. Gilan. 18. Lorestan 19. Mazanderan 20. Markazi 21. Hamedan 22. Hormozgan 23. Yazd Weighted Composite Indices It must be kept in mind that the composite indices estimated in the above manner are faced with two difficulties; frrstly, only one agricultural variable has been taken as a representative of agricultural sector, which is hardly adequate. Secondly, as

42 lndustriali:ation.. and Regional Disparities pointed out by M.N Pal(1975), using pooled correlation matrix involving all variables from all specific sectors runs into some problems which are indicated below. i). most variables from different sectors have normally low intercorrelation, whereas the method of first Principal component is only useful when selected variable are highly correlated. Otherwise the first principal component is not able to explain a high percentage of the variation. ii). the indicators of economic development are not necessarily independent, while the principal components must, by very assumptions, be orthogonal, and iii). "The economic interpretation of each principal component which involves all variables is very subjective and difficult~ all principal components evolved from a heterogeneous group of variables need not always be economically meaningful. While, on the other hand, the variables of a specific group are so chosen as to measure the same factor or group-character and as such there is no ambiguity or absurdity in the interpretation of this factor". 16 For example, percentage of labour force engaged in industrial activities is taken as an indicator of industrial development. This variable is completely inversely related to agriculture sector, that is to say, Kuznets has taken the high share of agriculture sector of total labour force as an indicator of backwardness. Therefore, it cannot be considered to be appropriate to obtain a composite index reflecting both agricultural and non-agricultural development, by the first principal component alone 17. To get rid ofthe above mentioned problems, which may influence the estimated results, we shall construct composite indices by using a w~,;ighted linear combination of the index of agricultural development and the index of non-agricultural M.N. Pal, "regional Disparities in the level of Development" in India, Indian Journal of Regional ~ vol VII, No.1, p. 36, 1975

43 Industrialization... and Regional /Ji.~parities. I SO development. For tackling this, twenty one variables have been selected at two points oftime namely, 1972 and 1991, as follows: lnfrastructural & ljrbanization Variables I. No. of vocational/technical students per 100,000 of population, 2. No. of Secondary students = 3. No. of beds 4. No. of doctors 5. Urban population per sq. km. 6. Postal facilities per 100,000 population 7. Length of roads per sq. km. 8. Percentage of urban population. Industrial Variables 9. Value added per worker in manufacturing activities, 10. Percentage of employed workforce in industry sector, 11. Share of manufacturing in total industrial workforce, 12. Per capita value added in manufacturing activities. 13. Average wages and salaries of manufacturing workers, 14. No. of manufacturing workers per one thousand sq. km. 15. No, of factories per one million population, Agricultural Variables 16. Distributed chemical fertilizer per acre, 17. Irrigated area per agricultural worker, 18. Percentage of irrigated area out of total cultivated area. Tertiary Variables 19. No. of labour engaged in tertiary sector per sq. km. 20. Percentage of labour employed in the tertiary sector, 21. No. of labour employed in the sector per population. After calculating the composite indices of provinces m the each subsection (infrastructure & urbanization, Industry and tertiary), they will be taken as variables to construct an index of non-agricultural development. The index of a!:,tticultural development, which is the first principal component of the sector, will also be considered as a variable to evaluated final composite index in manner, which had originally been adopted by M.N Pal (1975), already mentioned. The index of non-agricultural development (Y) IS the first Principal component of variables, Y i, Ti and li for 1971 where: Yl Index of Industrial developme-nt, Tl = Index of tertiary development li""lndex of infrastnlctural denlopment 17 Despite this, we have formed pooled correlation matrixes in the previous section in order to construct the composite indices on account of unavailability to sufficient data in the period under consideration.

44 Industrialization.. and Regional IJi"f){Jr/tie.\ lsi The highest eigenvalue is which explains %93.2, a very high percentage out of total variation. The corresponding correlation matrix is given below: CORRELATION MATRIX (1991) Ii Yi Ti li 1.00 Yi Ti The corresponding factor matrix is presented as follows: Factor 1 li Yi Ti The factor loadings of first principal component reveal that it has a positive significant correlation with all of them at 1% of probability. As usual, we form four categories in order to arrange the provinces based on their composite indices in terms of non-agricultural variables as follows: TABLE 5.16 Categories of Provinces ( 1971) Province Composite index (Score) A. Devel!.med &r!.!!!l! I. Markazi & Teheran.l B. Relativelr ~~veiqu~~ &rouu I. Yazd Esfahan Khuzestan Gilan Khonuan SellliW Fan Mazanderan East Azar baijan ~. L~s d~v~lone~ K!J!UU t. Kermanshah ~ Hamedan ~ \'tl e5t A-za\"'bo~q n -o 2.7"r?l 4. Kerman ~ Slstan & Ba1uchestan ~ Lorestan ~ Chaharmohai & Bakhtiari ~ Bushehr & Hormozgan ~ Kordestan ~ Ham ~ Zanjan ~.975 D. l!nderdeveloued &rqul! l. Kuhklluyeh & Bolnthmad Rank s ll 12 t>

45 Industrialization... and ReKional Disparities IRAN ADMINISTRATIVE DIVISION IIY I'HC )\ "1!'\CE 1 'N2 Composite Index of Development ( 1971) INOF.X: Developed Relatively De-.eloped Less Developed Map 5.1

46 Industrialization... and HeKionai/Jiv>arities. 153 The index of non-agricultural development Y I is the first principal component of variables Yj, Tj and lj for 1991, where: YJ lndu trlal denlopment Index Tj lj the Index of tertiary development the Index of infra~tructural development The highest eigenvalue is 2.19 which explains 73 percent out of total variation. Although this percentage is quite high, but it is less than that of The. corresponding correlation matrix and factor matrix are given below: "CORRELATION MATRIX FOR (1991)" Tj Yj lj 1.00 Yj Tj " 0.~905* "Factor matrix 1991" lj Yj Tj Factor I 0.881* 0.755* 0.920* Tj 1.00 Significant at I% level The provinces have been placed in the four classes according to obtained scores as follows: TABLE 5.17 Categories of Provinces (1991) Province Composite index d Develooed a.oue. I. Markazl & Teheran lj., f!.ellllivelv tlew/oi!.d fl!!!.!!l!. l. Yazd Esfahan Khuzestan Gilan Khoruan Semnan Fan Mazanderan East Azar baijan ~ /.Q!I da:,eloe!.!l. fl!!!.ui! l. Kermanshah HAmed an Z8 3. West Azarbaijan Kerman Slstan & Baluehestan Lorestan Chahannohal & 811khtlari Bushchr & Honnozpn Kordestan llam I:. Zanjlll'! D. fl.nder!kj!.elofl.r.d fl!!!.!h!. 1. Kubklluyeh & Boirahmad Rank s 9 10 II significant at % 5 probability 19 significant at% I probability

47 Industrialization... and Regional DivJOritie.L 154. IRAN ADMINISTRATIVE DIVISION OY PROVINCE: 1992 Composite Index of Development ( 1991) INOEX: Developed Relatively Developed Less Developed ~~"" Underdeveloped Map 5.2

48 lnduslria/i::ation... and Regional Disparities Comparison of scores obtained by the provmces m non-agricultural group reveals that Gilan emerged as a developed province in terms of the selected indicators in On the other hand, Khorasan falls in less developed one in Finally, Sistan & Baluchestan lost its place and got classified in underdeveloped group whereas, Kuhkiluyeh & Boirahmad held a higher rank and moved upwards to the less-developed group from underdeveloped one in General Composite Index In order to work out a general composite index, following M.N. Pal's work, a weighted linear combination from agricultural development index and non-agricultural development index will be used. The formulated M.N. Pal's formula is given below: Z = o 1 (WI X)+ o 2 (W 2 Y) Where Z composite index of economic development, 0 1 productivity per worker in agricultural sector at national level divided by Productivity per worker at the whole of the economy, l)z productivity per worker in non-agricultural sector at national level divided by productivity per worker at the whole of the economy, W 1 and W 2 stand for the proportions of agricultural and non-agricultural workforce respectively. The weights are the share of agricultural and non-agricultural sectors m Gross National Product of Iran (G.N.P.) in terms of productivity of agricultural and non-agricultural workforce divided by productivity of the total labour in the country, and percentage of labour employed in agricultural and non-agricultural sectors. It should be noted that, in absence of data, we must use national level value added per worker in agricultural and non-agricultural sectors instead of their productivity. We shall compute the CO!nposite index (Z) for 1971 and 1991 separately as follows: o, = national lenl value added per ~ orker in agriculture nationaj lnel value added pu worker In the total nationalle\'el value added per worker in non-agriculture national level value added per worker in the total

49 !ndustriali=ation... and Regional Disparities.. I 56 The calculated values of 8 1 and 82 are given as follows : Yrar ~ ~ The formulae are as given below for 1971 and 1991 respectively (W X) (W Y) (WI X I (W 2 Y 1 ) Table 5.18 The Position of Provinces Based on General Composite Index Composite Rank Composite Province Index (Z, Index (Z, 1971) 1991) Eut Axarbaijan West Azaroaijan Esfahan llam Bushehr & Hormozgan Chaharmoha1 & Bakhtiari Khorasan Khuzestan ZanJan Semnan Slstan & Baluchestan Fan Korde~tan Kerman kermanshah Kuhkiluyeh & Boirahmad Gllan Lorcstan Mazanderan Markazl & Tehran 4.75 I 3.18 Hamed an Yazd Rank I 18 5 The computed values of composite indices and the ranks of provinces are presented in the table No. 13. We have categorized provinces into the four categories, based on their composite indices. Results are shown in table No.14: A look at the above tables reveals that some provinces like Ham, Zanjan and Bushehr & Horrnozgan, which were located at the bottom of the hierarchy in 1971, improved their ranks in 1991 namely, Ilam moved from rank 22 to 16, Zanjan from rank 21 to 14 and Bushehr & Hormozgan from rank 20 to 10. On the other hand, the rank of Khorasan, Kordestan and Kuhkiluyeh and Boirahmad deteriorated noticeably. Markazi & Teheran, which had occupied the highest place in 1971, remained at the

50 !ndu5tria!izalion.. and Regional!Ji.,parities. 157 top category in 1991 as well. Esfahan has also remained at the same rank during the period under consideration. It is worth noting that Yazd lost its place in developed category in favour of Gilan, in Finally, it should be mentioned that provinces like, Kordestan, East Azarbaijan, Hamedan and Kermanshah have faced the Iraq-Iran war in period , and Khorasan was faced with bulk migration of Afghani migrants. Therefore, it may be said that outbreak oflraq-lran War and crowding in the migrants are the main reasons which explain their deterioration after the revolution. Table 5.19 Classification of Provinces Based on General Composite Index :-:=: :-::-::-:------[J OProvince Compodte Rank Province Composite Rank index(z) [J index (z) A. Developed group 1. Markazi & Tehran 2. Yazd 4.~ A.Developed group Markazi & Teh..J.l8 Gilan 1.30 I 2 B. Relatively developed grop 1. E1fahan 0.9~ 2. Gllan Semnan Mazanderan 0.31 ~. Khuzestan Khorudn Kermamhah Fan 0.09 c. Less developed group I. Kerman -0.0~ 2. Eut Azar West Azar Lorestan Slstan & Balu Hamedan Chaharmohal & Bak Kordestan Kuhldhayeh & Bolr Bushehr & Hor Zanjan !lam ~ B.Relatively developed group Esfahan Semnan Yazd 0.~3 ~ Khuzestan Kerman Ma7.anderan Fan C. Less developed group Bushehr & Hon West Aur II Kermanshah Slstan & DahL Zanjan East Azar I lam Loreatan Hamedan -0.6~ 18 Khorasan Chaharmohal & Bak D. Underdevelped group Kuhkiluyeh & Bolr Konlestan

51 Industrialization... and Regional Disparities Composite Index of Development 5 4 Comp. Ind Province ic Fig East Azarbaijan Fan 2. West "" Esfaban 4.Dam S. Busbebr & Hormozgan 6. Cbaharmobal & Bakb. 7. Kborasan 8. Kbuzestan 9. Zanjan 10.Semnan 11. Sistan & Balu. 12. Fan 13. Kordestan 14. Kerman IS. Kermansbab 16. Kubk:iluyeb & Boi. 17. Gilan 18. Lorestan 19. Mazanderan 20. Markazi & Tehran 21. Hamedan 22. Yazd