A Study of the Formal Manufacturing Sector in India: Performance of Significant Industries and Spatial Influence

Abstract

This chapter looks into the performance of the organised manufacturing sector by analysing technical efficiency (TE) at the firm level, in particular to test the impact of locational concentration of firms. A key objective of the study is to determine if such spatial concentration of firms within a state has any influence on the performance of those firms. The microeconometric analysis is duly controlled using expected determinants of TE including a wide array of state-specific infrastructure parameters. It covers all significant manufacturing industries that together contribute over 80% of the total value added of the formal manufacturing sector. The study observes, quite expectedly, that the size of firms has a significant positive contribution to TE for all of the major industries analysed. Government-owned firms are seen to be less efficient compared to their privately owned counterparts. The interesting aspect of the findings is in the variance observed in the impact of spatial concentration on efficiency. In general, firms located in the urban districts are more efficient. Locational concentration of firms is found to have a positive contribution on performance for sectors like  automobiles, coke oven products and machinery and equipment. However, surprisingly, a high degree of spatial concentration is seen to have a negative effect on efficiency for some key sectors like basic metals, food products and beverages, the chemical sector and pharmaceuticals. We attribute this finding to the diseconomies emanating from congestion, higher prices and higher wages as undesirable effects of high industrial concentration, which practically outweigh the positive economies expected from greater access to infrastructure, technology, skill base and knowledge spillovers in the industrialised states.

Portions of the chapter is drawn from a paper presented by the authors at the IEA World Congress, Jordan, 2014. Comments from the session participants are gratefully acknowledged.

Appendix

1.1 Concordance Table of Industries

The study period includes ASI data sets which are based on three different industry classifications, i.e. NIC 98 used for the ASI microdata for 2000–01 and 2003–04, NIC 04 used for the period 2007–08 and NIC 08 used for 2011–12. Accordingly, an assessment of concordance has been carried out at the two-digit level between the NIC codes used in the input data. At the two-digit level, all the relevant industry codes have the same scope in both NIC 98 and NIC 04. But mapping between industry classifications at the two-digit level was required for NIC 04 and NIC 08 to make the 2011–12 file usable. Concordance and related mapping are presented below (Table 6.3).

Table 6.3 Industry mapping between NIC 04 and NIC 08 for comparability over time-series

1.2 Construction of the Infrastructure Index Using Principal Component Analysis

There are several dimensions of physical infrastructure development which are often correlated in the spatial context. In isolation, the key indicators of physical infrastructure may provide insights into firm performance. However, when the key indicators of infrastructure are used collectively, the analysis will undoubtedly be more robust. Using a large number of disaggregated indicators of infrastructure in the regression analysis is also constrained by the problem of multicollinearity. It is in this context that the statistical technique of PCA is applied in constructing a single index that captures the variance or information contained in several infrastructure variables and thus overcoming the problem of multicollinearity. The multivariate statistical technique of PCA finds linear combinations of the original variables to construct the principal components with a variance greater than any single original variable. In this study, we have used four major components of infrastructure of the Indian states, viz. (i) road: proportion of surfaced to total road in each state, (ii) telecommunication: state-wise density of telephone (per 100 population), (iii) credit-deposit ratio: state-wise distribution of credit and deposit by scheduled commercial banks of India (as a ratio), (iv) electrification—urban and electrification—rural: state-wise distribution of urban and rural households having electricity (in percentage).

While the first and second principal components are used for the regression analysis, the second component was later dropped due to strong correlations with the first component. The first component explaining roughly 50% of the variability has been used as an index of infrastructure within the pooled regression analysis (Table 6.4).

Table 6.4 Eigenvalues and cumulative proportions by principle component