REGIONAL IMBALANCES IN A.P. AN APPLICATION OF TAXONOMIC METHOD AT DISTRICT LEVEL - G. Lakshminarayana, M.A., M.Phil., Asst. Director, DE&S, Hyderabad (1996) SECTION I SETTING THE GROUND: Development
does
not
pervade
uniformly
and
perhaps
more
particularly in a democracy. To reverse it is the goal of every government. Economic planning has emerged as a powerful means to achieve this end. There however, are to serious flaws in the theory and practice of planning efforts that still exist not only in India but in developed economies also. It was expected that the cumulative benefits of economic planning would ‘trickle-down’ automatically or at best be handed down in an administrative fashion to the majority of masses. The results of Green Revolution for example spelt out the failure of this paradigm. The auspicious launch of green revolution package no doubt, resulted in the substantial increase of agricultural production. The other side of the coin shows widened economic and social inequalities between the rich and the poor (Krishna Rao, 1978; Hanumantha Rao, 1975). It is because the problems and prospects were not analyzed for regions having different natural, economic and social environs. Another distortion in Indian Planning is over emphasis on ‘time’ while undermining ‘space’. Out Five Year Plans put targets to be achieved in GDP growth rate, industrial production etc. the fruits of industrial development for instance have not been equitably distributed spatially which resulted in the concentration of industries and establishment of units without taking the needs, linkages and levels of development of different regions into consideration.
The foregoing analysis brings out the need of regional development based on regional planning to the notice of planners at various levels. True, development is holistic and one should think globally but it is equally important to act locally through regional planning at micro-level by initiating a ‘bottom-up’ process. This has become an important aspect in the frame work for what economists labeled as ‘Another Development’ (Haque etal, 1975). WHAT IS A REGION? It is necessary at this juncture to know the concept of a region. A region is a spatial unit having some contiguity and homogeneity. Hence, a region my be economic or physical (Learmount 1958), resource endowment (Sinha, 1968), metropolitan (Nath, 1972), macro-micro-meso region (Census, 1961), nodal or programme region (Sharma, 1966).
OBJECTIVES OF THE PRESENT STUDY:
The process of regional planning has various steps, viz., (i) identifying and demarcating the regions (ii) determining the needs of the region (iii) making a plan for the region (iv) implementing the plan and (v) monitoring and evaluation of the plan (Bhale Rao, 1968). The priorities and design of a regional plan will be determined basing on the present levels of development or backwardness of regions. The relative position of a region vis-à-vis other regions hence becomes increasingly important as this dimension discloses, though partially the pace and location of development. The present study makes an attempt in this direction. Further it compares the results with other studies.
UNIT OF ANALYSIS:
In the Indian context the district is chosen as the Unit of analysis by many-a-regional economic analyst. Considering the current status of
availability of disaggregated data the district seems to be a more convenient unit of analysis. The choice of a district as the unit of analysis does not preclude interpretation at the regional level as well (Prabhu and Sarker, 1992). SCOPE AND DATA BASE:
The exercise has been carried out for all the districts in the state except Hyderabad. The results will lose consistency if the Hyderabad District is included since it is purely an urban district. the remaining twenty two districts were considered for the study. Data pertaining to twenty one important variables covering demography, agriculture, industries, health, education
infrastructure,
labour
and
employment
and
transport
and
communication etc., have been collected for, the latest available year and used for the purpose. The data were collected from respective departments apart from D.E. & S., METHODOLOGY: The identification of backwardness or development can be made basing on different techniques. Earlier, problems involving mass data analysis such as regionalization were handled by overlaying maps. It was replaced by multivariate procedure with the help of sophisticated calculating machines. Factor analysis is such a multivariate procedure which reduce many variables to a few principal components representing ‘complexes’. The limitation of this method is that though the main component explains the maximum variance in the series the proportion of variance explained hardly exceeds 50 percent (Dasgupts, 1974; Chakravarthy Committee, 1981).
The other procedure is the ‘method of taxonomy’ through which we get optimal classification of observations into the regions on the basis of the distances between observations in the factor space (Berry& Ray, 1966). This is used in various ways, ranging from simple aggregation (Dandekar committee, 1989) to standardized raking of distance measure. The main
drawback of this method is the lack of an objective weight system resulting in equal weights to all indicators (Dholakia, 1985).
The present study has adopted the method of taxonomy in its developed from due to its substitutability.
A SURVEY OF EARLIER STUDIES IN A.P.: Apart from the Techno-Economic Survey (1956) by NCAER, the state Planning department attempted to identify backward areas over the years. Here is a chronological presentation. (I) 1962: To formulate special programmes for the development of chronically drought affected areas on the basis of minimum average annual rainfall in each taluk for a period of 21 Years between 1942-62, 76 taluks in the state have been identified as drought affected. (II) 1965: To distribute resources among Community Development Blocks more rationally, the blocks were classified into advanced, ordinary, backward and tribal basing on sis indicators covering agriculture, population and transpiration by assigning arbitrary weights to them. (III) 1974: The Government of India declared 14 districts in the state as industrially backward for concessions by Financial Institutions. The coverage was extended to neutralize the political unrest during 1973-74. (IV) The State Planning Department has worked out five exercises to identify backward areas. In the first exercise the ranking method was adopted. The other exercises were based on indices of development with state average taken as 100. The indicators chosen for these exercises covered population, agriculture and transportation. The State Planning Board, however, decided that the identification of backward areas should be done in each region.
(V) the Technical Committee under the chairmanship of BPR Vithal (1978) and the CESS (1981) also attempted to identify backward areas at taluk level using the statistical technique ‘Principal Component Analysis’. It is evident from the above description that nobody has so far attempted to apply the ‘Taxonomic Method’ to identify levels of development among the districts of A.P. SECTION –II This section delineates the proposed exercise adopting the method of taxonomy. The district of Hyderabad which is wholly urban is ignored to make the analysis meaningful. Twenty one indicators were chosen covering economic and social characteristics of districts. INDICATORS TAKEN FOR THE STUDY: DEMOGRAPHIC INDICATORS : 1.Percentage of Urban Population. II. AGRICULTURAL INDICATORS : 1.Gross irrigated area as a percentage of gross cropped area. 2.Net sown area as a percentage of cultivated area. 3.Fertiliser (N+P+K) consumption per hectare of gross cropped area. 4.Average size of operational holdings. III.
INDUSTRIAL AND LABOUR INDICATORS :
1.Percentage of other than agricultural labourers to total main workers. 2. work participation rate (General). 3.Number of factory workers per lakh population. IV.
INFRASTRUCTURAL INDICATORS :
1.Agricultural pump sets energized per 100 hectares of net area sown. 2.No. of tractors per 1,000 hectares of net area sown. 3.Percapita consumption of electricity (domestic).
4.Road length per 100 sq. kms of geographical area. V.
MONEY AND BANKING:
1.Percapita bank credit.. 2. No. of Cc-operative Societies (all-types)
VI.
SOCIAL INDICATORS:
1.Hospital beds per lakh population 2.Number of fair price shops per lakh population 3.Couple protection rate (CPR). 4.Literacy rate (general). 5.Drop-out rate (Classes 1-7) VII.
OTHERS:
1.No. of periodicals and newspapers per million population. 2.No. of telephone connections per lakh population. METHOD OF TAXONOMY: ‘Taxonomy’ is a term used particularly in biology to define population and assign new specimens to these classes which are called classification and identification respectively by taxonomists (Fisher, 1936). Taxonomic method was developed by polish Mathematicians in 1950s for non-biological purposes. It was used by United Nations (1968) and U.P. State Government earlier to identify the levels of development among units. Being simple and lucid to understand it is an improvement over simple ranking techniques. In the opinion of economists it can best be substituted for sophisticated techniques like factor analysis etc. (Mandal, 1988). Following is the description of the method of taxonomy. The method with statistical notations is given in the appendix. (I) SELECTION OF INDICATORS: The taxonomic method allows the indicators to be progressive and regressive types.
(II) STANDARDIZATION OF INDICATORS: Standardization makes indicators unit free and dimension less and hence aggregation becomes meaningful. The standardized value is obtained by subtracting the mean of each indicator from it and them dividing it by its Standard Deviation. (III) ASSUMING AN IDEAL REGION: The ideal region contains the highest values of positive indicators and the lowest values of negative indicators which are to be simulated from the regions taken for the exercise. Then standardized values of the ideal region have to be identified. (IV) PATTERN: The pattern for a region quantifies the distance of that region from the ideal region. (V) MEASURE: The measure indicates the level of development or backwardness ranging from 0 to 1. A value close to ‘O’ denotes the highest level of development and a value close to ‘1’ denotes the least development. We can express the measure in percentage terms by multiplying it with 100. THE RESULTS: The following table gives pattern and measure values and ranks of respective districts. Table I
SLNO.
DISTRICT
PATTERN
MEASURE
RANK
1
Srikakulam
14.25
0.812
19
2
Vizayanagaram
14.48
0.825
20
3
Vishakapatnam
10.88
0.62
4
4
East Godavari
11.96
0.681
10
5
West Godavari
11.19
0.637
6
6
Krishna
8.73
0.497
1
7
Guntur
10.94
0.612
3
8
Prakasam
13.81
0.786
17
9
Nellore
11.16
0.636
5
10
Kurnool
13.58
0.773
14
11
Ananthapur
13.68
0.779
15
12
Kadapa
13.37
0.761
13
13
Chittoor
11.33
0.645
7
14
Rangareddy
11.52
0.656
8
15
Nizamabad
10.59
0.603
2
16
Medak
13.71
0.781
16
17
Mahaboobnagar
15.69
0.894
22
18
Nalgonda
13.04
0.743
12
19
Warangal
11.78
0.671
9
20
Khammam
13.84
0.788
18
21
Karimnagar
11.99
0.683
11
22
Adilabad
14.79
0.842
21
Krishna District stands
first in the ranking order
followed by
Nizamabad, Guntur, Vishakapatnam, Nellore and West Godavari occupying first six ranks. Mahaboobnagar stands last in the ladder having Adilabad, Vizayanagaram, Srikakulam, Khammam and Prakasam districts preceding it. The remaining districts are of middle range. This group consists of Chittoor, Rangareddy, Warangal, East Godavari, Karimnagar, Nalgonda, Kadapa, Kurnool, Ananthapur and Medak. This paper concentrates on ‘how’ the picture is rather than to comment on ‘why’ it is so. The enquiries like the factors that helped Nizamabad to secure second position are the subsequent steps of this paper. Indicators chosen and method adopted may to some extent explain the ladder of ranking.
LIMITATIONS OF THE STUDY:
(i) Equal weight age is given to all the indicators (ii) The measure values of two time periods are not directly comparable. This may happen because of the change in the values of the ideal region though there is no change for a district. This exercise perhaps, is the maiden attempt to apply Taxonomic Method to identify the levels of development among the district in Andhra Pradesh. However, this method was applied among the mandals of Nalgonda district (Hanumantha Rao, 1987) earlier. SECTION-III This section attempts to compare other exercises done by two premier organization, viz., CMIE (Centre for Monitoring Indian Economy) and CMS (Centre for Media Studies) with the present one.
THE CMIE EXERCISE (1993) : CMIE ranked the districts in India taking national average as 100. simple aggregated ranking method was adopted. The weightage structure is as follows: I. AGRICULTURE SECTOR:
35%
a) Percapita volume of output of principal crops
25%
b) Percapita Bank Credit to Agriculture
10%
II. MANUFACTURING SECTOR:
25%
a) Workers per lakh population
15%
b) Percapita Bank Credit to Manufacturing Sector III. SERVICES SECTOR
10% 40%
a) Percapita Bank Deposits
15%
b) Percapita Credit to Services
15%
c) Literacy
4%
d) Urbanization
6%
With this framework, the districts were ranked. THE CMS EXERCISE (1993-93) :
The CMS, Delhi in collaboration with the Royal Netherlands Embassy studies Soci-Economic Development Status of A.P. Gujarat, Kerala and Uttar Pradesh. THE METHOD : There are two basic components, viz., Economic Development Value (EDV) and Social Development Value (SDV) having equal weightage of 0.50 each. The EDV contains percapita income (0.20), percentage of households having electricity (0.10), percentage of population employed as main workers (0.10), and percentage of households having no exclusive room or having one room only (0.10)). The SDV consists of age-specific enrolment ratio for the age group of 614 years (0.10), literacy rate (0.10), infant mortality rate (0.20) and percentage of households having sanitation facilities (0.10). ESTIMATION PROCEDURE: For each indicator the district having highest value is given ‘1’ and adjusted values for other districts are calculated and these values for each variable are multiplied with respective weights. The EDV and SDV were computed by aggregation and these two values were added to arrive at development index and the districts were ranked accordingly. The following table gives a comparative picture of rankings of the three methods. TABLE -2 RANKING OF DISTRICTS
RANK 1 2 3 4
D.E. & S. Krishna Nizamabad Guntur Vishakapatnam
CMIE Krishna West Godavari Guntur Nizamabad
CMS Rangareddy Nizamabad Guntur West Godavari
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Nellore West Godavari Chittoor Rangareddy Warangal East Godavari Karimnagar Nalgonda Kadapa Kurnool Ananthapur Medak Prakasam Khammam Srikakulam Vizayanagaram Adilabad Mahaboobnagar
Karimnagar Nellore Rangareddy East Godavari Vishakapatnam Ananthapur Medak Prakasam Chittoor Kadapa Kurnool Warangal Khammam Nalgonda Adilabad Vizayanagaram Srikakulam Mahaboobnagar
Krishna Ananthapur Medak East Godavari Nellore Kurnool Kadapa Nalgonda Chittoor Khammam Prakasam Mahaboobnagar Warangal Vishakapatnam Karimnagar Vizayanagaram Adilabad Srikakulam
It is seen from the above table that the CMS ranking differs from CMIE and D.E. &S rankings. This deviation however is not significant as seen from the co-efficients of rank correlation. I.CO-EFFICIENT OF RANK CORRELATION: Rank Correlation Co-efficient are computed using the formula: P=1-(6 Sum(di2)/n(n2-1) Where, D= Difference in ranks between paired items in the two series. N=No. of paired observations. P=co-efficient of rank correlation. The co-efficients are: (I) DE&S and CMIE =0.83 (II) DE&S and CMS =0.67 Hence, the two methods of ranking are highly correlated with the present method of ranking. II TESTING OF HYPOTHESIS:
We may verify/test whether these three sets of ranks are dependent using ‘t’ test. The Null Hypothesis (Ho) : The two sets are independent. Maintained Hypothesis (Hm) : Two sets are not independent.
(i.e.,
dependent) We can compute ‘t’ value from the formula: ‘t’ = p-o/ Where, p=Co-efficient of rank Correlation. =Standard error of ‘p’, i.e, (
)
The critical (table) and computed ‘t’ values are given below. If the critical value is less than the computed value, we may reject the null hypothesis. Critical ‘t’ Value
computed ‘t’ value
(i) DE&S and CMIE
2.89
6.64
(ii) DE&S and CMS
2.89
4.04
(* degrees of freedom = 20) The computed ‘t’ values for both the sets are higher than the critical (table) ‘t’ values at 1% level of significance and hence the dependency of two sets is maintained. III. CO-EFFICIENT OF CONCORDANCE : We further attempt to know the extent of concordance between DE&S method of ranking and other two methods. W=125/M2[N(n2-1)]
where,
W= Co-efficient of concordance S=Sum of squared deviation of the actual sums from their Mean M=Total no. of rankings N= No. of items (* ‘W’ varies between 0 and 1) The Co-efficient of concordance among the three methods is 0.81. Higher level of concordance is obvious.
CONCLUDING REMARKS: Many methods of regionalization or clustering have been proposed and some of them proved useful, but none is thoroughly understood and established firmly. It is very hard to find out all satisfactory and objective sets of weights and indicators. It is also true that the results obtained by multivariate approaches could be had by univariate approaches just as easily (Ghosh). Only through experience and comparative experimentation can be understand how and why procedures fail or work.