Shifting patterns in developmental rank ordering: a case study of the districts of Sind province.

Jamal, Haroon ; Malik, Salman

The primary objective of this paper is to observe the changing patterns in regional development and to highlight some of the major underlying phenomena. Our examination of changes in rank ordering over a short period of time reveals that only moderately developed districts have altered their position in either upward or downward direction. Larkana district moved from sixth to fourth rank, while Tharparkar district regressed considerably from fourth to eighth rank. The exercise will facilitate policy-makers in allocating development resources in districts where the deficiencies are evidently serious. It will also be helpful in locating research areas to determine particular bottlenecks to development in districts which shifted downward in rank ordering.

I. INTRODUCTION

The decade of the 70s has witnessed efforts by various national and provincial governments in Pakistan to reduce regional disparities in the level of development. A number of explicit spatial policies have been pursued towards this objective. These have included, on the one hand, a range of fiscal and monetary incentives (1) to industrial investment in backward areas and mandatory restrictions, (2) on the other hand, on the establishment of new units in certain industries in the developed metropolitan regions of the country like Karachi and Lahore. In addition, there has been a dimension of regional equity in the allocation of public expenditures on physical and social infrastructure. The changed sectoral emphasis, with higher priority on rural development, is likely also to have had some favourable consequences on the spatial distribution of national income and welfare.

The basic question that arises is the extent of progress that has been achieved as a result of these policies in reducing regional imbalances in the level of development at the district level in Pakistan. It is the objective of this paper to identify the shifting pattern in the development rank ordering over the 70s in one province of the country, viz, Sind. This is a province which has traditionally been characterized by fairly sharp differences among districts in the level of development with the largest city of the country, Karachi, dominating economic activity in the province and the rest of the province being essentially rural and relatively underdeveloped, except for a few pockets of development in districts like Hyderabad and Sukkur.

The basic conceptual view of development adopted in this paper is to see it as the end result of interaction between various technological, economic, social and institutional factors. As such a series of indicators have been used to indirectly measure territorial levels of development [see Torgerson, Warren S. (1958)]. These indicators primarily relate to development inputs and can be used as tools for regional planning.

The paper has been organized as follows. Section II describes the particular indicators used for evolving the spatial and inter-temporal ranking of the districts of Sind in terms of level of development. Section HI discusses the methodology used for combining the various indicators into one summary measure of the level of development. Section IV presents the results of the empirical analysis. In Section V, we highlight the major implications of the findings and indicate some of the limitations of the exercise. Finally, the Appendix to the paper describes the various data sources for the different indicators and includes the background tables to the study.

II. CHOICE OF DEVELOPMENT INDICATORS

A number of studies have been undertaken in the past to rank districts of Pakistan in terms of level of development. Helbock and Naqavi (1976) used a multidimensional approach to determine the relative level of development at the district level for the 60s. Pasha and Hasan (1984) have attempted a similar analysis for the 70s. Khan and Iqbal (1982) have used the most recent data from the Agricultural Census of 1980, to quantify the spatial variation in the level of development of rural areas of the country in terms of access of villages to basic inputs and services.

All the above studies have, however, constructed development rankings at one moment in time. In this paper, given a consistent set of development indicators and data sources, an attempt is made to show how districts have changed their rank orderings over a period of ten years, from the early 70s to the early 80s.

Indicators that have been included in the study relate to measures of economic potential and achieved levels of income and wealth; mechanization and modernization of the rural sector, especially agriculture; housing standards and access to basic residential services; development of transport and communications; and availability of health and educational facilities. The individual indicators chosen are described below. Sources of data for the different indicators are described in Appendix (Table A-1).

Income and Wealth

Various indicators have been used for deriving the income and wealth position of a district. These indicators correspond to the traditional measures of the level of development. For the rural economy, agricultural value added per capita (AGVAD) is the basic indicator. It includes value added in major and minor crops, fishing and forestry. The method of computing the value added is, more or less, the same as adopted in national income accounts. Another important indicator of income and wealth in rural areas is live-stock per capita (LSTOCK). Different types of livestock have been aggregated by assigning weights. The extent of commercialization of agriculture (CCROPS) has also been used as a proxy for the relative modernization and prosperity of a rural area. It represents the proportion of total cropped area being used for the cultivation of cash crops like cotton, sugar-cane, rice, etc.

For the urban component of a district, per capita manufacturing value added (MVAD) has been used as an indicator of income levels. Due to the paucity of data, only value added in the large-scale industrial sector has been considered. Therefore, this measure may create distortions in the case of districts which have a relatively high presence of small-scale establishments.

An overall measure, albeit crude, of wealth levels included in the study is the number of bank branches per 10,000 persons (Banks). The assumption is that there is a direct correlation between the number of branches and the volume of bank deposits in a district. This assumption may at least partially be violated in the case of districts which are geographically large and access to banking services for the dispersed population may imply, other things being equal, a bigger network of branches.

Modernization of Agriculture

There is some debate as to whether mechanization of agriculture confers net social benefits. On the one hand, it contributes to higher yields and greater farm income leading to a rise in the standard of living while, on the other hand, there is evidence that the use of tractors and other farm implements had led to tenant evictions, labour displacement and more skewed distribution of land holdings. However, an indicator of the extent of mechanization of agriculture in the form of tractors per 1000 acres of cropped area (TRACT} has been used in this paper to capture the process of modernization in the countryside. In addition, the extent of use of fertilizer, measured as consumption or fertilizer as proportion of cropped area (FERT), and access to canal irrigation systems and tube-wells, in terms of irrigated area as proportion of cropped area (IRRI), have also been used as indicators of modernization in farming practices.

Housing Conditions

Improvement in housing conditions is one of the important consequences of socio-economic development. Possible indicators in this sector include the proportion of households without homes, the proportion in 'Pucca' dwellings, access to toilets by type, water connections, electricity and gas, etc. Again availability of data has restricted the choice to three indicators, viz, proportion of households using electricity (ELECT), gas (GAS) and with inside water connection (WATER)

Communications

Number of Radio sets (RADIO) per 1000 population, Television sets (TV) and number of Post offices (POST) per 100,000 population have been taken as indicators of the access to media and development of communication facilities. Data on number of residential telephone connections (TELEP)were not directly available, hence these have been estimated in a manner described in the Appendix (Table A-4).

Development of the transportation network of a district has been seen in terms of metalled road mileage (MROAD) and unmetalled road mileage (UMROAD) per 100 square miles of geographical area. With regard to the availability of transport vehicles, the use of a summary measure, viz., passenger load carrying capacity per capita, was first suggested by Pasha and Hasan (1984). The same approach, with some modification, has been used in this paper. The aggregate measure has been broken up into two indicators primarily to indicate availability of transport for lower and upper income groups in a district. For the former, by using appropriate weights, the number of scooters, motorcycles, taxis, autorickshaws and buses per 1000 urban population (PASSEN) has been determined. For the latter, the measure is cars, jeeps and station wagons per 1000 urban population (CARS).

Health

A number of indicators of health conditions are discussed in the literature. However, district-wise statistics on indicators such as life expectancy at birth, infant mortality rate, maternal mortality rate, etc. are currently not available. Hence, indicators of inputs rather than outputs of health services were used to explain the development in this sector. These are, first, the number of beds in hospitals, dispensaries and rural health centres (BEDS) and, second, the number of doctors per 10,000 population (DOCTORS).

Education

One of the basic indicators of development is the literacy rate. However, this indicator could not be included in the analysis because of the difference in the definition of literacy in the population censuses of 1972 and 1981. Therefore, instead of a stock measure of the development of the education sector, indicators of flow of output were used. Gross enrollment as proportion of population in the relevant age group are defined for primary (PENR), middle level (MENR), higher secondary (HENR) and intermediate and degree college level (IDENR).

Further Teacher-school and Teacher-student ratios have been included to depict the quality of education. In both types of ratios, three levels viz., Primary (PTSC and PTST), Secondary and Matric (HTSC and HTST) and Inter Degree (IDTSC and IDTST) have been considered.

Altogether, the total number of indicators included in the study is 31. There are five indicators of income and wealth, three of modernization of agriculture, three of housing conditions, eight of transport and communications, two of health and ten of education. (3)

III. METHODOLOGY FOR RANKING DISTRICTS

The simplest numerical procedure in establishing ranks of districts is the summation across indicators of standardized scores in each indicator. This technique is referred to as the Z-sum technique, and is described as follows:

[(Z-sum).sub.j] = [n.summation over (i=1)] [[X.sub.ij] - [[bar.x].sub.i]/[S.sub.i]] (1)

where

n = Number of indicators;

[[bar.x].sub.i] = Mean value of ith indicator;

[S.sub.i] = Standard deviation of ith indicator; and

[X.sub.ij] = Value of ith indicator in jth district.

The higher the Z-sum for a particular district the more developed it is in relation to other districts. The basic problem with this technique is that it assigns equal weights to all indicators.

The concept of Taxonomic distance is also used for the purpose of ranking of countries or territorial units on the basis of selected indicators. This procedure establishes the difference between an 'ideal' unit and the observed unit. For our purpose, an 'ideal' district is one which has 'best' value for a particular indicator. After standardizing indicators, taxonomic distance is obtained using the following formula:

[(TD).sub.j] = [[[n.summation over (i=1)] [([Z.sub.ij] - [[??].sub.i]).sup.2]].sup.1/2] (2)

where

[Z.sub.ij] = Standardized value of ith indicator in jth district; and

[[??].sub.i] = Highest standardized value of ith indicator among all districts;

The most developed district is one which has the minimum taxonomic distance. This technique also attaches equal weights to all indicators. In addition, the contribution to TD by a particular indicator can be increased if its maximum value, [[??].sub.ij], is very large in relation to the values for other districts. Therefore, this technique is sensitive to the presence of outliers. (4) In fact, in the context of Sind, Karachi district has very large magnitudes for some indicators in relation to other districts. Therefore, this district has been dropped from the analysis in order to avoid biases in the derivation of taxonomic distances.

Another common and popular multivariate method for indexing level of socioeconomic development is the technique of factor analysis. (5) This technique proceeds by clustering indicators which are correlated most into factors such that the latter is a linear combination of the former. In addition, it attempts to create factors which have minimum correlation among each other. Weights are assigned to each factor on the basis of eigen values, and in this manner the overall factor score for a district is computed.

Helbock and Naqavi (1976) and Pasha and Hasan (1984) have both used the Z-sum technique and factor analysis for ranking districts. Khan and Iqbal (1982) have, however, relied on the approach of taxonomic distances. In this study, the relatively simple techniques of Z-sum and taxonomic distances have both been used. The two procedures lead to almost identical ranks for the different districts in both time periods, thereby demonstrating the robustness of the results.

Due to the nature of the data set, factor analysis could not be used for ranking the districts. Bum (1982) has shown that spurious results may be obtained if the number of indicators is equal to or greater than the number of spatial units in the analysis. The reason for this is fairly obvious. If there are more variables than observations, the relevant co-variance matrix will be singular and its inverse will not exist. Zerby and Khan (1984) have provided an empirical illustration of this problem. In this study, with 31 indicators and ten districts only, the factor analysis technique could not, therefore, be applied.

IV. EMPIRICAL FINDINGS

Table 1 gives the magnitude of the Z-sum and Taxonomic distance in 1971-72 and 1980-81 for the ten districts of Shad included in the study. There appears to be considerable stability in the rankings of the most and least developed districts. Hyderabad, Sukkur and Khairpur have remained the three top districts of Sind (excluding Kaxachi) while Jacobabad and Thatta continue to be the most backward districts. There has been some change in ranking only among the intermediate districts. Larkana has demonstrated the greatest improvement followed by Dadu. On the other hand, Tharparkar has slipped considerably in its rank. Nawabshah and Sanghar have retained their relative position.

The robustness of the results is confirmed (6) by the fact that the correlation in ranks according to the two techniques is very high in both years. In 1971-72 the rank correlation coefficient was 0.99 and in 1980-81, 0.96.

Another important conclusion from Table 1 is that despite the various policy measures adopted, as mentioned earlier, regional disparities in the province of Sind have not tended to decline in the 70s. The evidence, at best, is ambiguous with respect to the change in the gap between the most and least developed districts. This is revealed by examining the change in taxonomic distances for each district over the period. Taxonomic distances have increased (implying reduction in the gap) somewhat for the two most developed districts, Hyderabad and Sukkur. However, they have increased significantly (implying increase in gap) for the two most backward districts, Thatta and Jacobabad. It is clear that a significant breakthrough has not yet been achieved in bringing the various districts of Sind closer to each other in terms of the level of development. (7)

To answer the question of regional disparities, explicitly, it is hypothesized that change in Z-score in each district for each variable is a function of base year's magnitude of Z-score. Using this regression specification, (8) 31 coefficients are estimated. The sign, magnitude and level of significance of [beta]'s are used to explain the changes in regional disparities over the time.

The important conclusions from this exercise are, first, all publicly provided municipal services--electricity, water and gas--have shown increase in disparity between most privileged and less privileged districts of Sind. This assertion is obtained from positive and significant coefficients (see Appendix, Table A-5). Second, inverse and significant coefficient of unmetalled road indicates that disparity has decreased in this development activity. Third, all negative coefficients in education has confirmed a general success of Government in terms of decreasing disparity among districts of Sind especially in primary education. Fourth, disparity among districts in modernization and mechanization of agriculture has, by and large, not altered. Fifth, it seems that disparities have decreased, to some extent, in per capita physicians availability. Finally, there appears no change in disparity in manufacturing value added. Thus it implies that fiscal and monetary incentives have not worked, properly, to decrease regional disparity in industrial investment.

Table 2 presents the sectoral ranks of each district in the two periods, it is possible from this table to determine the reasons for an apparent decline or improvement in a district's position. For example, the significant improvement in Larkana's status during the 70s can be attributed to a dramatic improvement (seventh to third) in the productivity of the district due partly to the rapid pace of mechanization and modernization of its agriculture (ninth to fifth). Simultaneously, there has been considerable expansion in social infrastructure, especially health (ninth to second).

The major deterioration in Tharparkar's position is due, first, to relatively slow growth in income and wealth (fifth to eighth) probably caused largely by the complete reversal (first to tenth) of the position of the district with respect to the use of agricultural inputs. (9) In addition, Tharparkar has witnessed a worsening of its rank in housing conditions (sixth to ninth) and health services (first to eighth).

Table 3 provides average sectoral growth rates for each district. An examination of data reveals that high growth in Dadu and Thatta is, mainly, because of substantive increase in per capita manufacturing value added. Significant improvement in Larkana's position in modernization and mechanization of agriculture and in health is the result of highest growth rates in both sectors. However, the district of Larkana has shown a decline in transport and communications facilities. In District Sukkur high growth in housing conditions and manufacturing value added per capita compensates for a significant decline in the sector of communications and health. High growth rate in Khairpur district (second highest) is due, first, to major improvement in road facilities (especially in unmetalled roads) and relatively greater improvement in housing conditions. The district of Nawabshah, Sanghar and Hyderabad have shown a moderate growth in virtually all sectors, while Tharparkar and Jacobabad have remained at the bottom in most development activities.

It is also possible from Table 3 to get an indication of which sector demands priority in future development allocations. All indicators of education have shown negative or relatively negligible growth in each district of Sind. In health, the situation is perhaps, even worse. All districts have shown significant declines in the availability of doctors with the exception of Larkana. Transport and road facilities also have shown retardation with a few exception. It is clear that development activities have not kept pace, by and large, with the growth of population.

Table 4 gives the rank correlation coefficients between sectoral ranks of the districts and the overall ranking for the two periods. A key conclusion is the relatively low correlation between income and wealth rankings and the overall status in terms of the level of development. This is consistent with the results of other studies [see Pasha and Hasan (1984)] and indicates that the use of per capita income only can distort the picture of development significantly. The highest correlation with the overall ranking is that of the housing, education and transport and communications indications. Therefore, the profile of backwardness in Sind appears to consist of a poor quality of housing, limited [UNREADABLE IN ORIGINAL SOURCE] to municipal services (water supply, gas, electricity) and restricted availability of health, educational and transport (roads) facilities.

V. CONCLUSIONS

The principal conclusion of the study is that despite the regional development policies pursued in the province of Sind during the 70s little success has been achieved in narrowing regional disparities among districts. This indicates that there is need for a fundamental re-evaluation of the nature, scope and content of these policies. Results of the study also indicate the priority sectors on which resources may have to be concentrated in future if the backward areas within the province are to be brought closer to the developed districts.

Appendices

Appendix Table A-1
Sources of Data for Development Indicators

 Nature of Data
Indicators Sources Manipulation

Income and Wealth

1. AGVAD 1. Development For detailed
 Statistics of Sind discussion of
 estimation procedure
 2. Markets and and computational
 Prices framework, see
 "Regional Accounts of
 3. District Census Sind 1970-71 to
 Reports, 1972 & 1981 1979-80" by Nuzhat
 Ahmad and Haroon
 Jamal, Applied
 Economics Research
 Centre (Unpublished).

2. LSTOCK 1. Pakistan Census This indicator is
 of Agriculture constructed using the
 (Province Reports), method adopted in
 1972 and 1980 Pasha and Hasan
 (1982). The weights
 2. Development assigned to each
 Statistics of Sind category of livestock
 are as follows:
 3. District Census
 Reports 1972 & 1981 Buffalo Cow & Camel 1.0
 Cattle 0.8
 Goats and Sheep 0.1
 Poultry 0.01

3. CCROP 1. Pakistan Census Rice, Sugar-cane,
 of Agriculture Cotton Rapeseed &
 (Province Reports) Mustard and Tobacco
 are considered as
 2. Development cash crops.
 Statistics of Sind

4. MVAD 1. Census of District-wise
 Manufacturing manufacturing value
 Industries added is obtained
 directly from the
 2. District Census appropriate census of
 Reports 1972 & 1981 manufacturing
 Industries.

5. BANKS 1. Banking Statistics Directly available.
 of Pakistan 1971-72

 and 1980-81

 2. District Census
 Reports 1972 & 1981

Modernization of
Agriculture

6. IRRI 1. Development Directly available.

7. FERT Statistics of Sind

8. TRACT 2. Pakistan Census
 of Agriculture
 (Province Reports)

Housing Conditions

9. ELECT 1. Pasha and Hasan Direct district-wise
 (1984) data on these
 indicators are
10. WATER 2. District-wise available for the year
 Housing Census 1980-81 in district-
 Reports (1981) wise Housing Census
 reports but an
11. GAS estimate was needed
 for the year 1971-72.
 Fortunately Pasha and
 Hasan (1984) estimated
 using Population
 Census figures of
 1971-72 and Housing,
 Economic and
 Demographic Survey,
 1973. For detail see
 Pasha and Hasan (1984,
 p. 186).

Communications

12. RADIO 1. Development Directly available.
 Statistics of Sind

13. TV 2. District-wise
 Census Reports

14. POST

15. TELEP 1. Pasha and Hasan The direct data on
 (1984) number of residential
 telephone connections
 2. District-wise were not available,
 Census Reports, hence an estimate was
 1972 and 1981 needed. For the year
 1971-72 the related
 figure were obtained
 from Pasha and Hasan
 (1984, p. 190).
 District-wise Census
 reports of 1981
 provide the total
 number of telephone
 connections including
 manufacturing concerns,
 therefore, to avoid
 double counting, these
 figures were adjusted
 by allowing 5 to 10
 percent deduction,
 keeping manufacturing
 value added of
 districts in view.

16. MROAD 1. District-wise Directly available.
 Census Reports, 1972
17. UMROAD and 1981

18. PASSEN 1. Development Directly available.
 Statistics of Sind

19. CARS 2. District-wise
 Census Reports

Health

20. BEDS 1. Development Directly available
 Statistics of Sind

 2. District-wise
 Census Reports

21. DOCTORS 1. Pasha and Hasan No published data on
 (1984) number of doctors are
 available. For the
 2. District-wise year 1971-72 number
 Census Reports of doctors per 10,000
 of population was
 obtained from Pasha
 and Hasan (1984).
 District-wise number
 of doctors for the
 year 1982 are
 available in an
 unpublished paper by
 N. A. Abbasi "An
 Analytical Evaluation
 of Personal Health
 Care Services,
 Organization and
 Management: Sind
 Regional Plan
 Organization".
 Karachi, 1983.

Education

22. PENR 1. Development All data are obtained
 Statistics of Sind directly. Enrollment
23. MENR ratio are calculated
 2. District-wise as percentage of the
24. HENR Census Reports following age groups:

25. IDENR 5-9 years for PENR

26. PTSC 10-14 years for MENR

27. PTST 15-19 years for HENR

28. HTSC 20-24 years for IDENR

29. HTST

30. IDTSC

31. IDTST

Appendix Table A-2
Taxonomic Distance by Districts

 Khairpur Jacobabad Sukkur

Total 1971-72 12.72 14.37 10.89
 1980-81 12.40 15.13 10.74
Income and 1971-72 4.87 3.04 2.82
Wealth 1980-81 4.73 5.65 4.59

Modernization 1971-72 2.31 4.79 3.45
of Agriculture 1980-81 1.25 4.15 3.60

Housing 1971-72 5.38 5.28 3.23
Conditions 1980-81 4.57 5.02 2.05

Communications 1971-72 7.62 7.56 5.30
 1980-81 6.80 7.70 5.00

Health 1971-72 3.98 4.24 3.75
 1980-81 4.08 4.53 4.26

Education 1971-72 5.48 8.34 6.78
 1980-81 6.86 8.45 5.85

 Nawabshah Larkana Sanghar

Total 1971-72 13.70 13.42 13.26
 1980-81 13.60 12.73 13.00
Income and 1971-72 5.18 4.35 3.58
Wealth 1980-81 5.32 4.64 3.68

Modernization 1971-72 2.11 4.34 2.30
of Agriculture 1980-81 2.33 3.30 2.09

Housing 1971-72 5.60 5.12 4.16
Conditions 1980-81 4.58 4.70 4.59

Communications 1971-72 7.10 6.30 7.50
 1980-81 7.30 6.50 6.76

Health 1971-72 4.71 4.70 4.48
 1980-81 4.55 2.98 4.46

Education 1971-72 7.22 7.36 8.02
 1980-81 7.58 7.35 7.97

 Tharparkar Dadu

Total 1971-72 13.12 13.79
 1980-81 14.14 13.50
Income and 1971-72 3.48 5.17
Wealth 1980-81 5.39 5.52

Modernization 1971-72 0.54 3.95
of Agriculture 1980-81 5.03 4.14

Housing 1971-72 5.28 5.56
Conditions 1980-81 5.40 4.99

Communications 1971-72 6.90 7.50
 1980-81 6.70 7.20

Health 1971-72 4.14 3.27
 1980-81 4.48 3.54

Education 1971-72 8.23 6.97
 1980-81 7.12 6.70

 Hyderabad Thatta

Total 8.08 15.15
 8.11 15.30
Income and 3.04 3.38
Wealth 4.69 4.72

Modernization 0.86 2.94
of Agriculture 2.13 3.82

Housing 0.00 6.15
Conditions 0.47 5.84

Communications 3.90 8.30
 2.10 8.60

Health 0.0 4.23
 0.0 3.57

Education 6.33 8.81
 5.88 8.59

Appendix Table A-3
Z-sum Score by Districts

 Khairpur Jacobabad Sukkur

Total 1971-72 2.19 -10.65 10.72
 1980-81 2.68 -16.77 15.06

Income and 1971-72 -1.12 3.11 1.71
Wealth 1980-81 0.04 -2.26 1.53

Modernization 1971-72 1.15 -4.28 -1.49
of Agriculture 1980-81 3.48 -2.26 -0.47

Housing 1971-72 -1.42 -1.23 2.35
Conditions 1980-81 -0.70 -1.47 4.40

Communications 1971-72 -3.31 -2.4 5.6
 1980-81 -1.2 -5.1 4.4

Health 1971-72 -0.32 -0.74 -0.01
 1980-81 -0.62 -1.22 -0.89

Education 1971-72 7.21 -5.09 2.62
 1980-81 1.72 -4.46 6.11

 Nawabshah Larkana Sanghar

Total 1971-72 -7.78 -6.51 -4.21
 1980-81 -5.94 -1.56 0.04

Income and 1971-72 -3.38 -1.69 0.74
Wealth 1980-81 -1.78 -0.17 3.07

Modernization 1971-72 1.44 -3.12 1.44
of Agriculture 1980-81 2.40 -0.79 2.34

Housing 1971-72 -1.79 -0.95 0.84
Conditions 1980-81 -0.74 -0.86 -0.77

Communications 1971-72 -1.5 .6 -3.1
 1980-81 -1.9 .2 0.0

Health 1971-72 -1.41 -1.32 -1.08
 1980-81 -1.27 1.41 -1.17

Education 1971-72 -1.09 -0.05 -2.81
 1980-81 -2.66 -1.36 -3.45

 Tharparkar Dadu

Total 1971-72 3.35 -8.30
 1980-81 -9.05 -6.54

Income and 1971-72 1.95 -2.92
Wealth 1980-81 -0.66 -1.26

Modernization 1971-72 3.76 -2.34
of Agriculture 1980-81 -3.47 -1.88

Housing 1971-72 -1.22 -1.73
Conditions 1980-81 -2.18 -1.43

Communications 1971-72 2.4 -3.1
 1980-81 -0.9 -2.3

Health 1971-72 -0.52 9.88
 1980-81 -1.22 0.28

Education 1971-72 -3.01 0.95
 1980-81 -0.64 0.06

 Hyderabad Thatta

Total 1971-72 35.23 -14.04
 1980-81 34.68 -12.60

Income and 1971-72 2.42 -0.83
Wealth 1980-81 0.79 0.69

Modernization 1971-72 3.21 0.25
of Agriculture 1980-81 -1.62 -0.96

Housing 1971-72 7.90 -2.75
Conditions 1980-81 6.70 -2.94

Communications 1971-72 11.0 -5.8
 1980-81 13.7 -6.9

Health 1971-72 5.25 -0.72
 1980-81 5.12 -0.42

Education 1971-72 5.48 -4.22
 1980-81 6.74 2.06

Appendix Table A-4
Mean and-Standard Deviation of Development Indicators

 Indicators 1971-72 1980-81

I. Income and Wealth

 1. AGVAD 331.00 834.33
 (Agricultural Value Added in (75.80) (355.88)
 Constant Rupees per Rural Person)

 2. LSTOCK 0.50 0.43
 (Equivalent Number of Livestock per (0.07) (0.07)
 Rural Person)

 3. CCROP 35.87 42.20
 [Extent of Commercialization of (6.67) (11.67)
 Agriculture (Percent)]

 4. MVAD 230.68 1861.18
 (Manufacturing Value Added in (107.31) (2133.70)
 Constant Rupees per Capita-Urban)

 5. BANKS 3.11 5.77
 (Bank Branches per 100 Thousand of (0.68) (0.96)
 Population)

 II. Modernization of Agriculture

 6. IRRI 84.26 83.06
 (Irrigated Area as a Percent of (15.04) (18.25)
 Cropped Area)

 7. FERT 0.01 0.02
 (Use of Fertilizer in Nutrient (0.01) (0.01)
 Tonnes per Acre of Cropped Area)

 8. TRACT 0.38 1.40
 (Tractors per 1000 Acre of Cropped (0.24) (0.55)
 Area)

III. Housing Conditions

 9. ELECT 10.71 25.09
 (Percentage of Dwelling Units with (4.18) (15.32)
 Electricity)

 10. GAS 0.91 4.54
 (Percentage of Houses with Gas (0.98) (4.99)
 Connection)

 11. WATER 5.59 10.50
 (Percentage of Dwelling Units with (3.42) (6.66)
 Inside Water Connection)

 IV. Communications

 12. RADIO 13.67 15.27
 (Radio Sets per Thousand (9.21) (8.04)
 Population)

 13. TV 6.20 303.15
 (Television Set per 100 Thousand (11.89) (248.21)
 Population)

 14. POST 7.26 6.09
 (Post Offices per 100 Thousand (5.41) (3.94)
 Population)

 15. TELEP 4.21 1.83
 (Residential Telephone Connection (1.80) (0.79)
 per 1000 Population)

 16. MROAD 3.20 6.64
 (Miles of Metalled Roads per 100 (1.81) (3.80)
 Sq. Mile of District Area)

 17. UM ROAD 8.34 11.77
 (Miles of Unmetalled Roads per 100 (7.10) (4.69)
 Sq. Mile of District Area)

 18. PASSEN 30.64 24.71
 (Passenger Load Carrying Capacity (24.82) (30.86)
 per 1000 Urban Population)

 19. CARS 2.16 1.79
 (Cars and Jeeps per 1000 Urban (1.24) (1.73)
 Population)

 V. Health

 20. BEDS 2.66 5.06
 (Hospital Beds per 10 Thousand (1.58) (2.51)
 Population)

 21. DOCTORS 1.01 0.69
 (Doctors per 10 Thousand (0.58) (0.45)
 Population)

 VI. Education

 22. PENR 35.00 32.00
 [Primary Enrollment Rate (8.00) (5.00)
 (percent)]

 23. MENR 2.00 2.00
 [Middle Enrollment Rate (Percent)] (1.00) (1.00)

 24. HENR 9.00 13.00
 [Higher Secondary or Matric (3.00) (3.00)
 Enrollment Rate (Percent)]

 25. IDENR 1.00 1.00
 [Inter-degree Enrollment Rate (0.10) (1.00)
 (Percent)]

 26. PTSC 2.31 2.59
 (Primary Teacher-school Ratio) (0.40) (0.44)

 27. PTST 0.04 0.04
 (Primary Teacher-student Ratio) (0.01) (0.00)

 28. HTSC 14.60 16.92
 (Higher Secondary and Middle (1.45) (1.64)
 Teacher-school Ratio)

 29. HTST 0.06 0.04
 (Higher Secondary and Middle (0.03) (0.01)
 Teacher-student Ratio)

 30. IDTSC 17.49 18.04
 (Inter-degree Teacher-school (3.46) (2.73)
 Ratio)

 31. IDTST 0.09 0.05
 (Inter-degree Teacher-student (0.06) (0.02)
 Ratio)

Appendix Table A-5

Results of Regression Analysis *

Indicators [beta]- t-Statistics
 Coefficient

AGVAD 0.15 0.09
LSTOCK -0.92 -2.86
CCROP -0.68 -1.12
MVAD 7.06 1.09
BANKS -0.72 -1.49
IRRI -0.65 -1.58
FERT 0.23 0.85
TRACT -0.59 -0.74
ELECT 1.92 2.46
WATER 0.88 4.93
GAS 3.90 7.75
RADIO -0.46 -1.89
TV 16.04 3.76
POST -1.16 -4.63
TELEP -0.62 -7.93
MROAD 0.43 0.79
UMROAD -0.73 -3.43
PASSEN -0.23 -0.67
CARS -0.31 -0.73
BEDS 0.28 0.84
DOCTORS -0.30 -2.58
PENR -0.46 -2.88
MENR -0.25 -0.82
HENR -0.41 -1.57
IDENR -0.34 -1.50
PTSC -0.02 -0.08
PTST -0.79 -4.64
MTSC -0.56 -1.51
MTST -0.94 -1236
IDTSC -0.69 -2.70
IDTST -0.86 -11.75

Note: * Specification: [Z.sub.it+1] - [Z.sub.it] = [alpha]
+ [beta] Zit Where; [Z.sub.it] = Z-score of district i in
period t.

Appendix Table A-6
Magnitude of Development Indicators for Districts of Sind *

 Khairpur Jacobabad Sukkur Nawabshah Larkana

 1971-72

AGVAD 342.870 459.850 383.630 220.020 326.650
LSTOCK 0.590 0.620 0.510 0.420 0.490
CCROP 21.910 37.260 34.080 33.940 41.370
MVAD 321.160 70.000 275.800 253.100 62.000
BANKS 2.250 3.810 3.540 2.580 2.600
IRRI 98.240 51.670 77.920 96.270 67.640
FERT 0.010 0.001 0.004 0.020 0.004
TRACT 0.490 0.130 0.320 0.310 0.070
ELECT 8.500 9.100 14.500 8.000 9.600
WATER 4.000 4.100 8.000 3.500 4.600
GAS 0.500 0.510 1.630 0.390 0.530
RADIO 16.001 9.200 16.964 6.998 12.052
TV 0.000 0.000 0.070 0.590 0.000
POST 8.080 2.120 21.810 5.980 6.510
TELEP 3.500 2.500 7.500 3.300 4.000
MROAD 1.620 3.080 2.310 3.430 3.900
UMROAD 0.640 5.120 10.970 17.630 15.000
PASSEN 5.350 23.650 48.450 19.300 39.740
CARS 2.370 3.740 2.900 1.420 1.990
BEDS 2.890 2.200 1.960 1.280 2.230
DOCTORS 0.740 0.750 1.260 0.700 0.400
PENR 40.000 31.000 49.000 27.000 40.000
MENR 4.000 2.000 2.000 2.000 2.000
HENR 11.000 7.000 4.000 8.000 10.000
IDENR 1.000 0.300 1.000 1.000 0.200
PTSC 2.650 2.130 2.560 2.570 2.160
PTST 0.050 0.030 0.030 0.040 0.040
HISC 18.290 13.930 15.160 15.790 14.590
HTST 0.050 0.060 0.140 0.050 0.040
IDTSC 14.800 13.330 18.000 17.000 17.330
IDTST 0.050 0.080 0.080 0.040 0.220

 1980-81

AGVAD 994.120 1119.270 425.070 741.460 856.490
LSTOCK 0.520 0.420 0.430 0.360 0.420
CCROP 27.700 44.660 62.210 39.240 50.580
MVAD 339.900 52.300 2041.800 1125.900 142.600
BANKS 6.010 3.660 6.620 5.990 5.780
IRRI 99.080 68.790 84.830 99.060 77.070
FERT 0.030 0.010 0.002 0.040 0.020
TRACT 2.600 1.050 2.000 1.320 1.340
ELECT 23.000 20.450 57.220 20.680 24.930
WATER 9.300 7.070 16.120 10.390 7.280
GAS 2.610 1.270 11.810 2.360 2.700
RADIO 14.165 10.726 29.200 6.175 11.264
TV 115.270 146.440 497.730 203.600 297.200
POST 2.240 1.380 1.220 3.540 11.310
TELEP 1.540 1.280 2.840 1.470 1.960
MROAD 10.290 4.050 4.870 10.910 8.580
UM ROAD 14.130 7.490 16.200 16.680 12.830
PASSEN 7.150 13.920 42.960 5.770 3.440
CARS 1.980 1.690 3.090 0.400 0.400
BEDS 4.260 3.820 2.470 3.640 8.510
DOCTORS 0.560 0.370 0.760 0.380 0.710
PENR 37.000 20.000 39.000 30.000 35.000
MENR 5.000 2.000 2.000 3.000 2.000
HENR 13.000 9.000 16.000 12.000 15.000
IDENR 1.000 1.000 2.000 1.000 1.000
PTSC 2.460 2.300 3.020 2.770 2.560
PTST 0.040 0.040 0.030 0.030 0.040
HTSC 18.400 17.440 19.370 16.660 16.590
HTST 0.040 0.050 0.040 0.040 0.040
IDTSC 14.670 14.600 23.800 18.750 15.800
IDTST 0.040 0.070 0.040 0.040 0.050

 Sangbar Tharparkar Dadu Hyderabad Thatta

 1971-72

AGVAD 321.400 422.220 209.040 340.570 283.770
LSTOCK 0.580 0.380 0.480 0.460 0.450
CCROP 35.320 41.020 28.610 38.140 47.080
MVAD 91.100 327.300 308.300 353.700 244.500
BANKS 3.900 3.540 2.610 4.030 2.200
IRRI 97.980 98.160 75.990 93.910 84.830
FERT 0.020 0.023 0.004 0.020 0.002
TRACT 0.240 0.640 0.140 0.770 0.680
ELECT 13.700 9.200 7.600 20.800 6.100
WATER 6.300 4.100 3.600 14.900 2.800
GAS 0.830 0.500 0.520 3.610 0.100
RADIO 5.560 33.744 7.766 25.645 2.761
TV 1.590 5.710 2.730 40.470 10.830
POST 9.380 2.850 8.440 2.930 4.540
TELEP 2.800 3.700 5.500 7.200 2.100
MROAD 4.690 0.900 1.980 7.590 2.470
UMROAD 2.440 0.500 8.850 20.700 1.490
PASSEN 17.480 86.790 3.800 52.140 9.750
CARS 0.740 3.980 0.350 3.260 0.810
BEDS 1.410 1.310 4.370 6.620 2.310
DOCTORS 0.840 1.200 0.890 2.590 0.720
PENR 37.000 23.000 40.000 36.000 24.000
MENR 1.000 2.000 1.000 3.000 1.000
HENR 13.000 9.000 8.000 14.000 3.000
IDENR 1.000 1.000 1.000 2.000 0.100
PTSC 2.120 1.600 2.440 3.020 1.840
PTST 0.040 0.060 0.040 0.040 0.050
HISC 13.790 13.370 14.250 13.430 13.360
HTST 0.050 0.050 0.050 0.050 0.090
IDTSC 13.250 17.600 24.670 22.400 16.500
IDTST 0,040 0.060 0.070 0.060 0.220

 1980-81

AGVAD 1701.980 782.710 530.760 699.790 491.670
LSTOCK 0.480 0.570 0.350 0.370 0.420
CCROP 40.480 18.000 43.450 50.400 45.240
MVAD 195.800 1335.800 5637.100 1521.300 6219.100
BANKS 6.620 5.590 4.830 7.200 5.350
IRRI 99.910 39.090 73.690 96.160 92.930
FERT 0.040 0.030 0.020 0.030 0.010
TRACT 1.470 0.680 0.690 1.640 1.250
ELECT 18.990 11.270 18.620 50.040 5.670
WATER 9.450 6.260 6.350 28.260 4.470
GAS 3.480 1.320 2.620 16.520 0.730
RADIO 17.796 18.031 6.896 29.656 8.787
TV 285.140 202.600 202.040 973.650 107.760
POST 9.110 10.320 11.520 3.990 6.290
TELEP 1.620 1.570 1.440 3.720 0.890
MROAD 5.690 1.270 3.170 13.730 3.900
UM ROAD 17.800 8.380 8.590 13.390 2.240
PASSEN 0.830 47.110 20.640 104.140 1.160
CARS 0.120 1.510 1.750 6.270 0.660
BEDS 3.390 2.860 6.210 10.660 4.810
DOCTORS 0.470 0.540 0.610 1.980 0.550
PENR 34.000 28.000 34.000 36.000 26.000
MENR 2.000 3.000 3.000 2.000 1.000
HENR 15.000 12.000 11.000 18.000 7.000
IDENR 1.000 1.000 1.000 2.000 0.300
PTSC 2.370 2.040 2.730 3.590 2.100
PTST 0.040 0.040 0.040 0.030 0.040
HTSC 14.790 17.000 16.000 19.010 14.000
HTST 0.040 0.040 0.040 0.040 0.060
IDTSC 17.000 21.000 17.670 19.600 17.500
IDTST 0.050 0.050 0.070 0.040 0.090

Note: * Excluding Karachi.

REFERENCES

Adelman, Irma, and C. T. Morris (1972). "The Measurement of Institutional Characteristics of Nations: Methodological Considerations". The Journal of Development Studies. Vol. 8, No. 3.

Bum, 8. (1982). "Factor Analysis and Development: A Note". Journal of Development Economics. Vol. 11, No. 1.

Helbock R.W., and S. H. H. Naqavi (1976). "Inter District Variation in Social Well-being in Pakistan". Islamabad: Pakistan Institute of Development Economics. (Unpublished Paper)

Khan M. H., and J. A. Zerby (1981). "The Socio-economic Position of Pakistan in the Third World". Pakistan Development Review. Vol. XXI, No. 3.

Khan M. H., and M. Iqbal (1982). "Socio-economic Indicators in Rural Pakistan: Some Evidence". Pakistan Development Review. Vol. XXI, No. 3.

Pasha, Hafiz A., and Tariq Hasan (1984). "Development Ranking of Districts of Pakistan". Pakistan Journal of Applied Economics. Vol. II, No. 2.

Torgerson, Warrven S. (1958). Theory and Methods of Scaling. Charlotte, NC: UMI Publications Ltd.

Zerby, J. A., and M. H. Khan (1984). "A Comparison of Multivariate Methods for Indexing Socio-economic Development". The Singapore Economic Review. Vol. 29, No. 1.

(1) To promote balanced regional development, general and/or region specific incentives are granted. For instance, complete exemption from customs duty is available on import of certain machinery and equipment in Shikarpur, Jacobabad, Tharparkar and Dadu (excluding Kotri). Industries in Hyderabad, Kotri and Karachi division, cannot avail this facility. In order to provide for timely credit in adequate quantities to different regions, mandatory credit targets have been given to the commercial banks. Tax holiday for industries in least developed areas is another example of region specific incentives.

(2) In Sind, there is a general ban on any new unit in Karachi, Dhabeji and Gharo, unless it can either only be established at these locations (ship-building and repairs, ship-breaking, sea sault, canning and preservation of sea food) or has downstream Links with major investments like Steel Mill (engineering workshops) or has a localized market (ice and cold storage, bakeries) or represents the application of complex, large-scale technology (petrochemicals).

(3) An exercise to verify the sensitivity of ranks to the number of indicators in each sector yielded almost similar rank ordering. We ranked districts after standardizing for the number of indicators using Z-score in each sector (for instance, Z-sum in income and welath divided by 5 plus Z-sum in Housing divided by 3 and so on). The rank correlation coefficient in overall Z-sum (standardized for number of indicators by sector)and overall Z-sum (without standardizing sectors) was 0.97 for the year 1980-81, while it was 0.92 for 1971-72.

(4) The technique of Z-sum is also sensitive (although less as compared with taxonomic distance) in case of outlier which effects the magnitude of the mean and standard deviation.

(5) For detailed discussion, [see Adelman and Morris (1972)].

(6) Because of this robustness, we will present the remaining results based on taxonomic distance technique.

(7) For the year 1971-72 the gap, according to mean taxonomic distances, between most and less developed districts emerges as 5.28 and for 1980-81 it comes to 5.75.

(8) The following specification is used:

[DELTA][Z.sub.ij] = [alpha] + [beta][Z.sub.ij]

where [Z.sub.ij]. is Z-score (with two periods being merged into one common data set) of district i for indicator j. Positive [beta] implies that regional disparity has increased, negative [beta] indicates reduced disparity and insignificant refers to the situation where no change in disparity among districts has observed.

(9) The coverage of cropped area for district Tharparkar in two censuses (1972 and 1981) is not the same. The latter was extended to some parts of Tharparkars (sub-division Mithi) which could not be covered in 1972 census. This extended area, largely, consists of unirrigated cropped area. Due to this extension, Tharparkar has shown a significant decline in the extent of irrigated area.

Haroon Jamal and Salman Malik *

* The authors are Research Economist and Research Officer, respectively, at the Applied Economics Research Centre, University of Karachi. They are grateful to Dr Hafiz A. Pasha, for his substantive comments and help in finalizing the draft of this paper. They also thank Sajjad Akhtar and two anonymous referees for their valuable comments. Any defects which remain are, of course, the sole responsibility of the authors.

Table 1
Magnitude of Z-sum and Taxonomic Distances

 1971-72

Districts Z-sum Rank Taxonomic Rank
 Distances

Khairpur 2.19 4 12.72 3
Jacobabad -10.65 9 14.37 9
Sukkur 10.72 2 10.89 2
Nawabshah -7.78 7 13.70 7
Larkana -6.51 6 13.42 6
Sanghar -4.21 5 13.26 5
Tharparkar 3.35 3 13.12 4
Dadu -8.30 8 13.79 8
Hyderabad 35.23 1 8.08 1
Thatta -14.04 10 15.15 10

 1980-81

Districts Z-sum Rank Taxonomic Rank
 Distances

Khairpur 2.68 3 12.40 3
Jacobabad -16.77 10 15.13 9
Sukkur 15.06 2 10.74 2
Nawabshah -5.94 6 13.60 7
Larkana -1.56 5 12.73 4
Sanghar -0.04 4 13.00 5
Tharparkar -9.05 8 14.14 8
Dadu -6.54 7 13.50 6
Hyderabad 34.68 1 8.11 1
Thatta -12.60 9 15.30 10

Table 2
Overall and Sectoral Rank Ordering by Districts
(According to Taxonomic Distance)

 Income and Modernization
 Overall Wealth of Agriculture

Districts 71-72 80-81 71-72 80-81 71-72 80-81

Khairpur 3 3 8 6 5 1
Jacobabad 9 9 3 10 10 9
Sukkur 2 2 1 2 7 6
Nawabshah 7 7 10 7 3 4
Larkana 6 4 7 3 9 5
Sanghar 5 5 6 1 4 2
Tharparkar 4 8 5 8 1 10
Dadu 8 6 9 9 8 8
Hyderabad 1 1 2 4 2 3
Thatta 10 10 4 5 6 7
 0.85 * 0.29 * 0.26 *

 Housing
 Conditions Communications

Districts 71-72 80-81 71-72 80-81

Khairpur 7 3 9 6
Jacobabad 5 8 7 9
Sukkur 2 2 2 2
Nawabshah 9 4 5 8
Larkana 4 6 3 3
Sanghar 3 5 6 5
Tharparkar 6 9 4 4
Dadu 8 7 8 7
Hyderabad 1 1 1 1
Thatta 10 10 10 10
 0.59 * 0.85 *

 Health Education

Districts 71-72 80-81 71-72 80-81

Khairpur 4 5 1 4
Jacobabad 7 9 9 9
Sukkur 3 6 3 1
Nawabshah 10 10 5 7
Larkana 9 2 6 6
Sanghar 8 7 7 8
Tharparkar 5 8 8 5
Dadu 2 3 4 3
Hyderabad 1 1 2 2
Thatta 6 4 10 10
 0.53 * 0.83 *

Note: * These figures are rank correlation coefficients between two
rank ordering in 71-72 and in 80-81.

Table 3
Average Sectoral Growth Rates by Districts

(Percent)

 Income Modernization
Districts Overall and of
 Wealth * Agriculture

Khairpur 4.2 0.0 13.6
Jacobabad 1.5 -3.8 21.2
Sukkur 2.9 5.5 6.2
Nawabshah 2.7 3.5 11.0
Larkana 3.9 1.6 22.0
Sanghar 2.1 1.5 11.8
Tharparkar 1.6 1.1 -2.9
Dadu 5.2 6.6 15.5
Hyderabad 2.8 1.9 5.7
Thatta 3.7 8.0 9.3

Districts Housing Communication **
 Condition

Khairpur 15.8 6.2
Jacobabad 9.9 -3.3
Sukkur 18.7 -3.8
Nawabshah 17.5 -5.2
Larkana 13.7 -6.2
Sanghar 9.7 -2.4
Tharparkar 7.0 3.2
Dadu 13.9 4.3
Hyderabad 13.6 1.9
Thatta 11.1 -1.5

Districts Health Education

Khairpur -0.9 -0.4
Jacobabad -4.6 0.3
Sukkur -4.5 2.0
Nawabshah -0.1 1.3
Larkana 16.8 1.1
Sanghar -1.1 1.1
Tharparkar -3.9 1.6
Dadu -2.2 0.8
Hyderabad -0.2 0.4
Thatta 1.5 1.6

Notes: * Agricultural and Manufacturing value added are in
constant prices.

** Excluded TV--Because of low or zero base in most of the districts.

Table 4
Rank Correlation Coefficients between Overall
and Sectoral Ranks

Sectors 1971-72 1980-81

Income and Wealth 0.32 0.58

Modernization of
Agriculture 0.50 0.64

Housing Conditions 0.71 0.90

Communications 0.67 0.83

Health 0.48 0.50

Education 0.72 0.79