5.1.1The economic activities pursued in any economy can be broadly classified into three sectors namely, primary sector, secondary sector and tertiary sector. The secondary sector, consisting of manufacturing, electricity and construction, plays a pivotal role in the economic development of the country. The percentage share of this sector in the nation’s GDP at current prices was about 23.9 per cent during 1999-2000. Within the secondary sector, manufacturing activities constitute a major share in the country’s GDP. During 1999-2000, of the total percentage share of 23.9 per cent in the GDP held by the secondary sector, manufacturing activities had a share of 15.4 per cent.
5.1.2 For collection of data relating to manufacturing and repairing activities, all units pursuing such activities in the country are grouped into two categories, namely, registered units and unregistered units. Hereafter ‘registered units’, unless otherwise stated, shall refer to the units registered under Sections 2m (i) and 2m (ii) of the Factories Act, 1948, or under the Bidi and Cigar Workers (Condition of Employment) Act, 1966, i.e. those employing 10 or more workers and using power and 20 or more workers but not using power. Within the manufacturing sector, registered units dominate over the unregistered units in the share of GDP. During 1999-2000, of the total percentage share of 15.4 per cent in the GDP held by the manufacturing units, the registered manufacturing units had a share of 10.0 per cent, the rest (i.e. 5.4 per cent) being accounted for by the unregistered manufacturing units.
5.1.3 Data on registered manufacturing and repairing units are collected through the Annual Survey of Industries and on unregistered manufacturing and repairing units, through the Follow-up Surveys of the Economic Census. This has been helpful in improving the degree of precision in the estimates for the manufacturing and repairing sector. For selection of units in the Annual Survey of Industries, the lists of factories maintained by the Chief Inspectors of Factories are used as the sampling frame. On the other hand, the Follow-up Enterprise Surveys (FuS), which adopt a stratified two or three-stage sampling design, with villages and urban blocks as the first-stage units (FSUs) and unregistered manufacturing and repairing units as the ultimate stage sampling units, generally use the list of villages and blocks with information on number of enterprises and workers as per the Economic Census as the sampling frame for selection of FSUs. Collection of data from the units in the Annual Survey of Industries is record-based. But in the Follow-up Enterprise Surveys, data are generally collected by interviewing the respondents, as most units do not maintain any books of accounts. The issues relating to the Annual Survey of Industries have been discussed in the succeeding paragraphs.
5.1.4 The Annual Survey of Industries is the principal source of Industrial Statistics in India. It plays a key role in assessing the changes in the growth and structure of the registered units in the manufacturing sector. Recognising the pivotal role of the registered manufacturing units, our development planners, as early as in the 1940s, felt the need for institutionalising the work of primary data collection, data processing and data dissemination pertaining to such units. This realisation resulted in the launching of a Census of Manufacturing Industries (CMI) and a Sample Survey of Manufacturing Industries (SSMI). The CMI and SSMI were in operation during the period 1944-58 and 1949-58, respectively. While the CMI was limited to factories registered under the Indian Factories Act, 1934, employing 20 or more workers, using power, and belonging to only 29 groups of industries; the SSMI was conducted with a wider coverage by including all the 63 industry groups in the country to meet the data requirements recommended by the National Income Committee set up in 1949. With the implementation of the Factories Act, 1948, the coverage of both the CMI and SSMI was extended to all factories employing 10 or more workers and using power, or 20 or more workers but not using power on any day of the preceding 12 months.
5.1.5 The Collection of Statistics (Central) Rules, 1959, framed under the Collection of Statistics Act, 1953, led to the launching of the comprehensive Annual Survey of Industries (ASI) in 1960 by replacing both the CMI and SSMI. Since then, this is being conducted regularly on an annual basis, except for 1972, when it was not conducted so as to help reduce the time lag in the release of results of the previous survey.Annexe 5.1gives the salient features of the sampling designs of ASI, 1973-74 to ASI, 1998-99.
5.1.6 The ASI covers the following categories of units:
All factories registered under sections 2m (i) and 2m (ii) of the Factories Act, 1948 employing 10 or more workers and using power, or 20 or more workers but not using power on any day of the preceding 12 months;
All bidi and cigar manufacturing establishments registered under the Bidi and Cigar Workers (Condition of Employment) Act, 1966, employing 10 or more workers using power, or 20 or more workers without using power;
Units engaged in certain services, repair of motor vehicles and a few other consumer durables like watches, etc. employing 10 or more workers using power, or 20 or more workers without using power; and
All electricity undertakings engaged in the generation, transmission and distribution of electricity that are registered with the Central Electricity Authority irrespective of their employment size as well as certain units engaged in activities of cold storage and water supply. From ASI 1998-99[1], however, all the units listed under this category are kept out of the coverage of ASI.
5.1.7 The units or factories in the ASI frame are grouped into census and sample sectors. While the factories in the census sector are surveyed on a complete enumeration basis, a representative sample from the sample sector is considered for survey in any survey year. During ASI, 1998-99, of the 1,71,675 units in the frame, a sample of 26,847 units – 9,008 from the census sector and 17,839 from the sample sector was actually surveyed. Annexe 5.2 gives the number of factories in the frame, the number allotted for survey and the number actually surveyed during ASI, 1993-94 to ASI, 1998-99.
Current Status
5.1.8 The ASI covers the entire country except the States of Arunachal Pradesh, Mizoram and Sikkim and the Union Territory of Lakshadweep. It is conducted annually under the statutory provisions of the Collection of Statistics Act, 1953, and the Rules framed there under in 1959 except in the State of Jammu and Kashmir where it is conducted under the State Collection of Statistics Act, 1961, and the Rules framed there under in 1964.
5.1.9 The Industrial Statistics Wing of the Central Statistical Organisation (CSO) of the Ministry of Statistics and Programme Implementation is responsible for the designing of the survey methodology, processing of data and preparation of reports. The work of carrying out the field operations for the survey rests with the Field Operations Division (FOD) of the National Sample Survey Organisation (NSSO). The Standing Committee on Industrial Statistics provides overall guidance regarding the survey methodology and tabulation of data.
5.1.10 Five States namely, Assam, Orissa, Rajasthan, Uttar Pradesh and West Bengal, supplement the ASI-related activity of FOD, NSSO, by covering a sample of residual units in the frame not covered by the FOD, NSSO, and canvassing the same ASI schedule in these units. The remaining State and Union Territory Governments (other than the States of Arunachal Pradesh, Mizoram, Sikkim and Union Territory of Lakshadweep, which are outside the coverage of ASI) do not survey any factories on their own. Each of the States and Union Territories (except the four States and Union Territories outside the coverage of ASI) receives copies of the schedules canvassed by the FOD, NSSO for their further processing or use.
5.1.11 Since 1998-99, the scope and coverage of the ASI has been limited to the registered manufacturing and repairing units only. The ASI frame is based on the lists of registered factories or units maintained by the Chief Inspector of Factories (CIF) in each State and those maintained by the licensing authorities in respect of bidi and cigar establishments.
5.1.12 The important aspects of the sampling design of ASI, 1999-2000 are as under:
All registered manufacturing and repairing units in the frame are grouped into two basic strata namely, the census sector and the sample sector. The units in the census stratum are surveyed on a complete enumeration basis.
The census sector comprises the following
All manufacturing and repairing units in the frame in the five States and Union Territories of Manipur, Meghalaya, Nagaland, Tripura and Andaman and Nicobar Islands; and
For the remaining States and Union Territories: All manufacturing and repairing units employing 200 or more workers.
In each of the States and Union Territories other than five States and Union Territories specified above, the complete list of units in the sample sector is stratified into different strata with each stratum consisting of all manufacturing and repairing units belonging to a particular industry 4-digit code of National Industrial Classification, 1998.
A sample of suitable size from each stratum is drawn circular systematically with equal probability and in the form of two independent sub-samples after arranging the units according to district and number of workers.
Of the total number of 1,74,167 units in the frame, the gross allotment of units considered for survey (which includes non-operating factories appearing in the frame) at the country level is 35,391 of which 9,570 are the census sector units.
5.1.13 Apart from information on name, address and some basic operational details about the factories, data collected in the ASI normally include the details of fixed assets; working capital and loans; employment details; input items consumed; products and by-products manufactured; other expenses and receipts as well as month-wise details of man-days worked, absenteeism and labour turnover for regular workers. Annexe 5.3 gives the items of information under the coverage of ASI, 1999-2000.
5.1.14 ASI is the major formal source for a comprehensive database on the different aspects of industrial activity in the country. Over the years, the entire work relating to the ASI has been streamlined. It has been the concern of all the agencies associated with the ASI survey work to take necessary steps so that it is possible to release the results within the shortest possible time. A number of measures, particularly those taken over a couple of years in the recent past, for speedy collection of primary data, quick data processing and data dissemination have led to the development of a system where the time lag in releasing the ASI results has been reduced considerably. As a result, it has been possible to prepare quick estimates of ASI 1999-2000 by the first week of January 2001. It may be mentioned that (see Annexe 5.4) the situation was not as satisfactory during the1980s and even the early 1990s. Detailed results for ASI 1983-84 were released only in December 1989, i.e., after a time lag of about four and half years and those for ASI 1989-90 and ASI, 1993-94 in January 1997 and April 1997, i.e., after a time lag of about six years and two years, respectively.
Select Issues
5.1.15 The Commission discussed the quality of ASI, data with a view to identifying remedial measures for improvement of the system. In this context, the following specific issues of concern were considered:
A. ASI frame
B. Accuracy of the estimates - Sampling and non-sampling errors
C. One-time census of industries
D. Schedule for data collection, computerisation and use of IT
E. Timeliness of processing
F. State participation in ASI
A. ASI Frame
Deficiencies
5.1.16 The lists of factories available with the States’ CIFs are used as a sampling frame for the ASI. The CIF lists, however, suffer from the following two types of deficiencies:
Many units that are eligible to be listed in the CIF list are not included in it; and
The units that have not been operating for quite some time are not removed from the CIF list.
5.1.17 Although it is claimed that there are built-in mechanisms for updating of the ASI frame every year, the available evidence clearly demonstrates that the ASI frame has remained grossly incomplete.
5.1.18 In order to assess the seriousness of the problem of non-inclusion of eligible units in the ASI frame, information on number of operating manufacturing and repairing enterprises estimated in ASI, 1994-95, estimated in NSS 51st round (1994-95) and counted in the Fourth Economic Census (EC, 1998) were analysed. The NSS considered only those units that were not included in the ASI. The units included in the 51st Round survey of the NSS, therefore, represent the units that are outside the ASI frame. The findings of the study, given in Annexe 5.5, reveal that:
In 1994-95, as estimated by the NSS 51st Round, about 1.45 lakh eligible units (i.e. employing 10 or more workers and using power or 20 or more workers but not using power) were not included in the ASI frame; and
Of these 1.45 lakh missing units, about 1.19 lakh units belonged to the employment size class 10 to 19 and the rest (i.e. about 0.26 lakh units) belonged to the employment size class 20 or more.
5.1.19 Further, the NSS estimate of total number of missing units does not include units that operated less than 30 days (15 days in case of seasonal enterprises) during the last year preceding the date of survey. The ASI, on the other hand, includes factories operating for any number of days during the last year. Thus, the number of units missing in the ASI frame is likely to be larger than 1.45 lakh. This huge magnitude of units missing from the ASI frame seems to raise serious doubt about the efficacy of the functioning of the CIFs in maintaining up-do-date registers of factories. Two conclusions follow: one, that the CIFs’ work of ensuring registration of factories by making use of the provisions that are available to them under the Factories Act is unsatisfactory; and two, that the matter requires a further study of the sample units which were eligible for registration but were, in fact, not registered.
5.1.20 The study also assessed the magnitude of non-included or missing units in the ASI frame not only in terms of number of units, but also in the corresponding number of workers and gross value added (Annexe 5.6). The findings are as under:
The ratio of the number of missing units (as estimated from the NSS 51st Round) to estimated number of working units as per ASI was of the order of about 117 per cent for the year 1994-95; and
The corresponding ratios with respect to the other two variables were about 28 per cent for total number of workers and 4.1 per cent for gross value added (GVA).
5.1.21 Thus the ratio of estimated number of missing units to the estimated number of working units as per ASI, as well as the corresponding ratio for the estimated number of workers, was quite substantial. But with respect to GVA, the ratio, say R, of total GVA by missing units to that by units covered under ASI was small (4.1 per cent).
5.1.22 Although the value of R is only 4.1 per cent at the all-India level, there are State-wise variations in it. The States and Union Territories for which the value of R exceeded the all-India ratio of 4.1 per cent are Delhi (14.5 per cent), Gujarat (6.2 per cent), Haryana (4.4 per cent), Meghalaya (8.3 per cent), Nagaland (8.5 per cent), Tamil Nadu (7.1 per cent), Uttar Pradesh (7.7 per cent), Andaman and Nicobar Islands (4.4 per cent), and Chandigarh (21.3 per cent). It is also quite likely that there will be considerable variation in the value of R for different industry groups.
5.1.23 Another deficiency of the lists maintained by the CIFs is that they include a large number of units, which have not been operating for quite some time. An examination of data of six ASI rounds, i.e., from ASI, 1993-94 to ASI, 1998-99 (Annexe 5.2) reveals this deficiency to be of the order of 15 to 20 per cent of the number of factories in the ASI frame.
Recommendations
5.1.24 Taking note of the inadequacies of the ASI frame, the Commission recommends the following:
The Chief Inspectors of Factories (CIFs) should update their own lists by including those units, which ought to be there in the lists and excluding the units that do not operate.
The CSO and the NSSO should take immediate steps to prepare directories of establishments (that employ at least 10 workers)[2] based on data collected in the Fourth Economic Census and the recent Follow-up Enterprise Surveys of Economic Census, respectively so that these could be made use of by the CIFs in their work of updating the lists maintained by them.
The work of preparation of a comprehensive frame for selection of samples in the ASI should be taken up jointly as a task on a priority basis by the NSSO and the CSO. This comprehensive frame should be prepared by taking into account the lists maintained by the CIFs, the directories of establishments prepared by the CSO and the NSSO and information available from other sources.
The comprehensive frame so prepared should be updated on a continuing basis using survey and census data as well as information available from other sources. The additional resource requirements for carrying out the work should also be met.
The State Level Coordination Committee (SLCC) should be reconstituted with the Secretary, Department of Industry, as the Chairman and the Labour Commissioner, Chief Inspector of Factories, Director of Industries, etc. as members so as to ensure greater vigour and discipline in the task of revision of the frames.
The CIFs should take steps to simplify the process of de-registration of closed units so that units non-operating for a long time could be removed from the registers of factories of the CIFs.
Accuracy of the Estimates - Sampling and Non-sampling Errors
Deficiencies
5.1.25 Collection of data in the ASI is based partly on the complete enumeration approach and partly on sample survey. As mentioned before, all factories in the census sector and a sample of those in the sample sector are covered for data collection. As the accuracy of various estimates made from the ASI would be affected by both sampling errors and non-sampling errors, a sound statistical system should continuously endeavour to reduce both kinds of error in the estimates. The following are a few important factors responsible for the sampling and non-sampling errors in the estimates:
Absence of a comprehensive list of factories
Limitation of sample size
Missing entries and typographical errors
Wrong calculations including totalling mistakes, and
Entries left blank due to transcription errors or other reasons.
Of these, the deficiencies in the list of factories have been discussed in detail before.
5.1.26 It is observed that during the last three years, there has been a drastic reduction in sample size. In this context, the Commission attempted to assess the precision of estimates of a few important survey characteristics based on the present sampling design and sample size, using recent survey data. The assessment was done for all industry groups combined, as well as for two- and three-digit industry groups at the all-India level. However, at the State and Union Territory level, the assessment was done for all industry groups combined and for two-digit industry groups (description of two-digit industry groups is in Annexe 5.7). Relative standard errors (RSEs) of the estimates of input, output and gross value added (GVA) were computed from the estimates of totals based on the two half-samples of ASI, 1997-98[3]. An analysis of the RSEs (see Annexes 5.8 to 5.12
) leads to the following conclusions:
For each of the three characteristics, the RSEs of the State and Union Territory-level estimates, as well as the all-India level estimates, for all industry groups combined were less than or equal to 5 per cent except for GVA in case of one State, for which RSE was found to be 6 per cent (see Annexe 5.8).
Of the 2-digit industry groups totalling 29, the number of industry groups with RSE of the all-India estimate greater than 5 per cent is 13 for input, 12 for output and 11 for GVA (see Annexe 5.9).
In 5 industry groups, the RSE of the all-India estimate exceeded 10 per cent for each of the three characteristics. Descriptions of the industries with RSE of an all-India estimate greater than 5 per cent along with the corresponding number of units in the frame and sample size (i.e. number of units surveyed) are given in Annexe 5.10.
The percentage of the 3-digit industry groups for which the RSE of all-India estimate exceeded 10 per cent for all three characteristics was about 41 to 42 per cent (see Annexe 5.11). In 42 to 44 per cent of the 4-digit industry groups, the RSE of the all-India estimate exceeded 10 per cent for all three characteristics.
In a large number of States and Union Territories, the RSEs of estimated output and GVA were above 10 per cent at the level of the 2-digit industry groups (Annexe 5.12). In the case of input, the picture was broadly similar. The situation would in all probability be worse at the 3 or 4- digit level of industry groups for States and Union Territories.
5.1.27 The Reports on ASI published by the CSO do not provide information on sampling errors. Interestingly, as long ago as in 1980, the Committee set up to review the national statistical system had recommended that statistical agencies should provide information on sampling errors along with the estimates based on sample surveys. This does not seem to have been implemented.
5.1.28 The Commission has noted various initiatives taken from time to time for reducing non-sampling errors. These include steps such as training of the personnel of the FOD (NSSO), computerisation of the data, adoption of measures for validation, etc. Despite these initiatives it is seen that non-sampling errors still creep in due to various reasons. Also a wide divergence is often observed of the summary or provisional results from the detailed or final results published afterwards. An examination of the estimates of gross value added based on the summary or provisional and detailed or final results for the five years, i.e. 1993-94 to 1997-98 (see Annexe 5.13) reveals that, except for ASI 1995-96, the divergence between the two varied from 3.6 per cent (ASI, 1994-95) to 16.0 per cent (ASI, 1993-94).
Recommendations
5.1.29 The Commission has studied the changes in the sampling design over the years, including changes in the sample sizes from time to time. It has also identified some of the major factors contributing to the non-sampling errors. After reviewing all these aspects, the Commission recommends that:
The effect of the changes made from time to time in the sampling design, including changes in sample size, on survey estimates should be analysed and the findings brought out regularly whenever such changes take place.
The sampling errors of estimates of important survey characteristics should be published along with the survey results.
The sampling design and the sample size should be reviewed to improve the precision of estimates at industry-group level. There is a need to increase the sample size for at least some of the industries where the sampling errors of the estimates are very high.
Non-sampling errors in the data should be regularly studied and measures should be taken to minimise them.
Factors responsible for wide divergence of summary or provisional results from detailed or final results need to be identified and steps taken to minimise such divergence.
C. One-time Census of Industries
Deficiencies
5.1.30 The major factors responsible for affecting the quality of ASI data are the deficiencies in the frame and the limitations imposed by the sample size on the task of generating disaggregated level estimates. It has been observed that the present sample design with the existing sample size results in fairly large sampling errors of estimates even at the all-India level for the main variables such as input, output and GVA, in some cases, even at the level of the 2-digit industry groups. Thus, the existing sample design of the ASI with its present sample size does not permit the generation of reasonably precise estimates for many industry groups covered in the survey.
Recommendations
5.1.31 Taking note of the various kinds of shortcomings described above, the Commission recommends the following:
A one-time census of industrial units, which are eligible for registration, should be conducted. While there should be a complete enumeration of all the units already included in the CIF list, the excluded units could be covered either by a sample or on a complete enumeration basis in the census. Such a census will help in generating reliable benchmark estimates at the disaggregated level (by geographical region, industry group, product level, etc.), providing an efficient weighting diagram for revision of the base year of the Index of Industrial Production, and also updating of the ASI frame
Before such a census is launched on a full scale, an exercise on a pilot basis should be carried out to explore the usefulness of the various alternative sources of information for preparing a comprehensive frame, and also to estimate resource requirements for conducting the one-time census.
D. Schedule for Data Collection, Computerisation and Use of IT Network
Deficiencies
5.1.32 It is often observed by the respondents that the contents of the schedule canvassed in the ASI do not conform to the pattern and the manner in which the data are maintained by the units. The respondents often argue that they can supply the data in a much better way when the demand for data conforms to the pattern in which they store the data. Further, the users of the data sometimes argue that the schedules do not collect some useful data that are required for analytical research. For example, some researchers have indicated out that the ASI collects information only on the written-down or depreciated value of capital, whereas information on the original value of capital when capital assets were procured by the undertaking would be more valuable. It is also pointed out that information on expenditure incurred for major repairs of capital assets, etc. could also be useful for analysis. Further, it is observed that the magnitude of the residual catch-all item ‘others’, in both the input and the output category, is often very large in the filled-in schedules and disaggregated information on this would be appreciated.
5.1.33 The Commission took note of the steps taken in computerising the data- processing work of the ASI. It noted, however, that no significant improvements have been made in collecting data from the industrial units through electronic media although about one third of the factories, according to the survey results, were found to use computers for accounting purposes during 1997-98 (Annexe 5.14).
Recommendations
5.1.34 The Commission recognised the need for rationalisation of the schedule. The Commission recommends that:
The schedule of data collection should be rationalised to make it as respondent-friendly as possible and designed in a computer-readable form.
The query,whether the industrial unit can supply ASI data on a floppy on software to be provided’ that formed part of the schedule of ASI 1997-98 but was subsequently dropped should be reintroduced. This would help in identifying units that are capable of furnishing information through floppy or other electronic media.
Information on input and output items clubbed under the large ‘others’ category should be obtained in a more disaggregated form through the schedule. Items having importance in value terms, at least, should be separated from ‘others’. This is particularly important for the census sector units.
For bringing about further improvement in the quality of data and a timely release of results, the cooperation of the factories should be solicited, through the Industry Associations and Chambers of Commerce, for supplying the required information in a pre-designed format in the electronic media.
Steps should be taken for data transmission from factories and field offices of NSSO to the tabulating agency in the form of digitised schedules using the countrywide IT network.
E. Timeliness of Processing
Recommendations
5.1.35 The Commission has noted (Annexe 5.4) that there has been a considerable improvement in the time frame for the release of ASI data during the last few years. The Commission recommends that:
Further measures should be taken to ensure the timely processing of data on a regular basis and release of the final results of ASI within one year of commencement of the fieldwork.
An advance calendar for release of quick estimates, provisional results, and final results should be declared through appropriate media, including a website.
F. State Participation in ASI
5.1.36 The Commission has noted that five States namely, Assam, Orissa, Rajasthan, Uttar Pradesh and West Bengal, are participating in the ASI by covering a sample of residual units in the frame not covered by the FOD, NSSO. It has also noted that all the States and Union Territories (other than the States of Arunachal Pradesh, Mizoram, Sikkim and the Union Territory of Lakshadweep, which are outside the coverage of ASI) are receiving copies of the schedules canvassed by the respective field offices of NSSO, to meet their specific requirements. The Commission has examined the mechanism of the five State Governments for collection and publication of the ASI data. It is seen that data collected by both the State and the Central agencies are used for tabulation by the respective State Governments.
Recommendations
5.1.37 As the kind of tables generated by the Central agency may not meet all the requirements of the States and Union Territories, the Commission recommends that:
The existing system of independent tabulation of ASI data by both the agencies should continue.
States and Union Territories should canvass the ASI schedules in the residual units (i.e., not covered by the NSSO) in their respective States and Union Territories, either on a sample or on a complete enumeration basis, whenever the required infrastructure and resources are available. For that purpose, the Government of India should declare, under the Collection of Statistics Act the Directors of the States’ Directorates of Economics and Statistics as statistical authority within the respective States, for the collection of ASI data and such other industrial statistics as the Government of India collects in its national programme.
