Abstract
The last decade (2001–2011) has witnessed a surge in the number of census towns (CTs) in India, which account for 30% of the country’s urban growth. Though several studies have tried to understand the spatial patterns and factors determining the emergence of these CTs, the all India level has been neglected. Due to an increase in non-farm activities, villages have been transformed into CTs. By considering 2,328 CTs at the all India level, this article investigates the relevant economic determinants of such transformation. To group similar CTs we use cluster analysis by considering several factors such as the size of the population of CTs, rural specific changes, climatic conditions, the growth dynamics of large cities which may spill over to rural hinterland, economic potential, the availability of infrastructures and job opportunities. The analysis suggests that the availability of infrastructure and the growth dynamics of the large cities are important for the emergence of these CTs, whereas rural poverty and unemployment rates do not seem to matter significantly. Finally, we suggest that for higher economic development, the rural to urban transformation is essential. For this purpose, the new CTs can offer an opportunity for increasing non-farm activities and the overall prospects for India. Hence, the policy directives will have to address the requirements of the CTs to emerge as centres of growth.
Introduction
The rural non-farm sector has shown its growth momentum in the last decade (2001–2011), resulting in the emergence of nearly 2,500 new urban settlements that are widely noted as census towns (CTs). 1 Despite their change in status, the local bodies in these towns remain under rural administration, not urban administration; even though their share of non-farming activities bypasses that of agriculture in terms of total employment. In other words, these towns are in transition; yet to be properly recognized as urban by the government of India, while census authorities refuse to categorize them as rural. Against the backdrop of this growth, questions relating to the determinants of such a transformation arise. After all, it is important to know the processes that initiated such rapid changes within a limited time horizon. Though this phenomenon was growing since independence, it was not so significant until at least 2001.
Referring to urban economics literature, one of the driving forces behind this shift may be owing to firms, which, in an attempt to take advantage of the benefits of agglomeration, choose to remain closer to the cities, though diseconomies do not permit them to invest within the city territory. Hence, the alternative for them is to exploit the rural space adjacent to the city boundaries. In light of this, the new census towns can be seen as a spillover of urban activities into the rural hinterland (Mitra & Kumar, 2015). On the other hand, the shift of labour to non-farm activities due to the lack of productive sources of livelihood in the agricultural sector is also a strong factor. This is especially the case since many census towns were seen to have emerged in the remote areas, far away from the large urban centres (Guin & Das, 2015; Roy & Pradhan, 2018).
Several studies (e.g., Chatterjee, 2014; Guin, 2018; Karmakar, 2015; Mukhopadhyay et al., 2016; Samanta, 2012; Sircar, 2016) argue that census towns have emerged due to structural transformations of the economy, including the decline of male workforce in agriculture. Pradhan (2013) and Mukhopadhyay (2017) have highlighted that this is due to the change in the economic structure of the existing settlements. However, most census towns are severely lacking in terms of an infrastructural base (Jain, 2018; Jain & Korzhenevych, 2020; Samanta, 2014).
In this study, we categorize new census towns (that have emerged as per the 2011 Census) based on cluster analysis. Our purpose is to identify the heterogeneity that may exist among this class of towns. These differences, in fact, may unravel the wide variations in mechanisms responsible for the growth and emergence of such urban centres and their performance indicators. These, in turn, may feed into the policy requirement, which can be pluralistic in approach.
Methodology and Analysis of Results
Several factors are hypothesized as being responsible for the emergence of new census towns. Cluster analysis helps us group similar units based on the observed values of several variables for each individual unit. This, in other words, allows us to identify sets of objects with similar characteristics. Though the K means clustering method is more efficient to handle big data sets; it requires prior knowledge of K, that is, the number of clusters we want to divide our data into. As we do not have any prior information regarding the number of clusters, we use the hierarchical cluster method for the analysis. This creates a series of models with cluster solutions from 1 (all cases in one cluster) to n (each case is an individual cluster). We follow agglomerative clustering in which most hierarchical methods fall into. We use Ward’s minimum variance method to specify a linkage algorithm to define the distance from a newly formed cluster to other clusters in the solution. The method combines those objects whose merger increases the overall cluster variance (i.e., the homogeneity of clusters) to the smallest possible degree. The approach is typically used in combination with (squared) Euclidean distances. The squared Euclidean distance increases the importance of a large distance over small distances.
It is very important to select clustering variables. We consider the following variables for categorization as follows:
The size of the population of census towns. Rural effects are measured by the total rural to urban migration, rural literacy rate, poverty headcount ratio and rural unemployment rate. As these factors are not available at the level of census towns, we consider them at the district level. A favourable climate may attract the rural population to a town. With the availability of data, we consider rainfall and temperature differences to capture the climate effect. Growth dynamics may spill over to the rural hinterland of the big city, which can generate new urban spaces or census towns. To capture the spillover effect, we consider the distance from the state headquarter, the nearby city with a population of 100,000 and above, and the nearest city with population 500,000 and more to a census town. We assume that the spillover effect declines with distance. Economic potential is also important for the census town. It is measured by the distance from a town to the nearest railway station. Better infrastructure of the town may attract people from rural areas. Infrastructure is measured by the town level: total road length, total number of latrines, total protected water supply, total number of electricity connections, total number of hospitals and total number of schools, colleges and universities. Job opportunities in the town are measured using the dummy variable, which considers 1 if a town has manufacturing industry and 0 if otherwise.
Table 1 presents the descriptive statistics of the variable used for cluster analysis. Data has been sourced from the town amenities, District Census Hand Book, Census of India 2011. However, as town level poverty cannot be calculated, we estimated it at the district level (i.e., the district to which the census town belongs) by using Rangarajan committee’s poverty line in 2011–2012 as well as the monthly per capita consumption expenditure based on the modified mixed reference period (MMRP) for the rural samples. In a similar way, the rural unemployment rate according to usual status (principal status + subsidiary status; adjusted), rural-to-urban migration rate and literacy rate are considered at the district level.
Descriptive Statistics
The pair-wise correlation coefficients of the variables used for the cluster analysis show that collinearity is not at a critical level. The variables such as log of town population and log of the total number of electricity connections show that the highest correlation is 0.74, which is clearly lower than 0.90 thresholds. This indicates that we can proceed to the analysis using all 18 clustering variables. 2
Following this, we decide the number of the clusters depending on statistical measures. Table 2 suggests that the largest Duda–Hart Je(2)/Je(1) stopping-rule value is 0.7243, corresponding to the 5th group. However, for this group, the pseudo-T-squared value is not the lowest and Calinski–Harabasz pseudo-F value is not the highest. Keeping this in mind, we consider 13-group solution with the second-largest Duda–Hart Je(2)/Je(1) stopping-rule value (0.7126) and lower pseudo-T-squared value (283.13) and a higher Calinski–Harabasz pseudo-F value (9407.8).
The Variance Ratio Criterion (VRC) and Duda–Hart Indices
The output in Table 3 shows that the cluster analysis assigned to all 1973 census towns unravels 13 segments. The first cluster comprises 890 towns (45%), the second cluster 704 towns (36%) and the fifth cluster 303 towns (15%). These are the three are the major ones among the 13 clusters. The rest of the clusters do not comprise more than 1% of the observational units each.
Number of Clusters
The mean values for the 13 clusters are given in Table 4. Comparing the mean values across the clusters, we find that among the different variables, the first cluster stresses literacy rate, amount of rainfall, road distance to the nearest city with a population of 5 lakh and more, road distance to the state headquarters, the total number of latrines and total protected water supply. The other variables hold comparatively less importance.
Comparison of Means
The main variables in the second cluster are rainfall, the total number of latrines, road distance to state headquarters and total protected water supply. The third and fourth clusters consider rainfall, road distance to state headquarters, the total number of latrines and total protected water supply. The fifth cluster stresses on rainfall, road distance to state headquarters, the total number of latrines, literacy rate and total protected water supply.
Other clusters also distinguish some of these variables as the predominant ones. These variables, that is, literacy rate, unemployment rate, rainfall, road distance to state headquarters, road distance to the city with a population of 1 lakh and more, road distance to cities which have a population 5 lakh and more, road distance to the nearest railway station, the total number of latrines and total protected water supply have, thus, play the most important role in clustering the new census towns into 13 segments.
Conclusion and Policy Implications
Our analysis of the census towns opens up a huge range of issues relating to the rural non-farm sector. Particularly, against the backdrop of the recent COVID-19 pandemic, it has been noted that a large number of migrants have moved back from large cities to their rural residences. Such processes of reverse migration are expected to raise unemployment and disguised unemployment in the rural areas. However, these situations of adversity can be converted to the advantage of the economy. The rural non-farm sector, as mentioned in the first section of the article, did not experience the growth of demand induced activities to any significant extent. It is now time to reflect on these considerations. The growth of small-scale industries in the rural areas can be encouraged as they would contribute significantly to rural growth and employment creation both. Even the large- and medium-sized industries, which are willing to benefit from the agglomeration process but cannot enter the territory of large cities due to entry barriers and other restrictions connected with city regulations, may be encouraged to invest in the rural areas adjacent to the large cities and the census towns. This will help them gain in terms of productivity, and the rural population on the other hand will be able to get absorbed in high productivity activities. The new towns will be able to contribute significantly to the economic growth and livelihood creation, both.
Overall, our analysis suggests that the emergence of the new towns is associated with infrastructure provisions and the growth dynamics of some of the large cities. In other words, new towns have been growing as satellite towns, with a strong relationship to the agglomeration economies that exist in large cities. Livelihood opportunities are explored with the provision of infrastructure endowment, which facilitates population movement from the rural areas to the hinterland of the big cities. While space in large cities tends to shrink with a concentration of economic activities, the rural hinterland provides land for new towns to emerge and benefit from agglomeration economies which remain within the reach of both the new firms and their workers. As a result, livelihood diversification becomes possible for the rural population, and earnings are expected to be higher as external economies of scale translate to higher levels of productivity.
So in terms of the policy implications, we suggest that the new census towns must be empowered with a higher level of infrastructure and job opportunities. The transformation from a rural agriculture-led economy to an industrial and service-led urban economy is an inevitable stage of development. The basic idea is that the urban economy uses resources such as land and labour more productively than the rural areas, hence, leading to higher economic development. Therefore, to facilitate the smooth transformation from rural to urban, the development of new census towns is very important. The proper management of new census towns will play a pivotal role in the context of higher and balanced urbanization (i.e., lower differences in population size among the cities and towns). This will also help reduce pressure on large cities and the impact of the other diseconomies, while the benefits of agglomeration economies originating from the large cities can still be attained. However, for this to happen in a significant manner, the census towns will have to be endowed with better infrastructure and investment in terms of urban facilities and services. The growth of demand-induced activities in the rural areas can actually contribute to strengthen the base of the new towns and the new urbanization processes that the country envisages to experience.
Footnotes
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.
Funding
The authors received no financial support for the research, authorship and/or publication of this article.