SAMPLE SURVEY OF UNREGISTERED SSI SECTOR

BACKGROUND

4.1

The Unregistered SSI sector was not surveyed earlier to the Third Census. The First and the Second Census of SSI units covered only the registered SSI units. The Economic Censuses so far conducted by the M/o Statistics & Programme Implementation also did not throw any light on the SSI sector. The Parliamentary Standing Committee on Industry in their 40th Report presented to Parliament on 2-5-2000 recommended that the number of units operating in the sector and the actual reason for having large number of unregistered units be ascertained.

4.2

The Steering Committee on Third Census of SSI units under the Chairmanship of Secretary (SSI) in its first meeting held on 27-10-2000 considered it feasible to cover the unregistered SSI sector through a sample survey along with the Third Census of registered SSI units. It was also decided in this meeting that the information on sickness and other characteristics to be collected in respect of registered SSI units would be collected for this sector as well.

4.3

The registration of SSI units is voluntary. The units, which satisfied the criteria laid down by the Central Government from time to time in terms of upper ceiling, in original value of plant & machinery (in case of SSIs and ancillary units) and in value of fixed assets (in case of SSSBEs) were registered at district level. These upper ceiling limits were policy driven and were always made applicable prospectively to new units seeking registration. The new units might not necessarily be newly established units. Some of the already established units might also have sought registration whenever the upper ceiling was enhanced, as they were not eligible earlier. Hence, it is not possible to state that the list of registered SSI units as on any date in the past bore the same classification in terms of the upper ceilings mentioned above. The same problems devolve to the unregistered SSI sector as well. This posed difficulties in defining the Unregistered SSI sector. Obviously, the definition changes with time. Hence, it has become necessary to fix a reference in terms of time and then prescribe a definition of Unregistered SSI sector.

DEFINITION OF UNREGISTERED SSI SECTOR

4.4

The Third Census on registered SSI units covered all SSI units permanently registered up to 31-3-2001 on complete enumeration basis. The Unregistered SSI sector for the purpose of Third Census has been defined as the set of all those units (SSIs, ancillaries & SSSBEs), which were eligible to be registered as on 31-3-2001, but were not permanently registered because the registration was voluntary. These included those units, which were temporarily registered on, or before 31-3-2001provided they were not permanently registered till 31-3-2001.

SAMPLING DESIGN

4.5

For preparing the sampling design for the survey on unregistered SSI sector, neither the sampling frame was available nor the area frame was available. The size of the sector was also not clearly known. In such cases, certain geographical units were to be selected and in those units the list of unregistered SSIs were to be prepared, so that a few of the units could be selected for survey. A statistical strategy had to be devised for picking up the geographical units and for selecting the SSIs to be surveyed. Statistical estimation requires prior knowledge of the population relevant for the survey at least at the level of the geographical units. The Economic Census 1998 gave information on number of units in the non-agricultural sector at village level and urban block level.

4.6

As no other source was available, the data on Economic Census 1998 furnished by the Central Statistical Organisation was used in formulating the sampling design for the unregistered SSI sector. The Economic Census (EC) data as per its coverage included the units of registered and unregistered SSI sectors, besides other units engaged in non-agricultural activities, which did not come under the purview of SSI sector. Efforts were made to eliminate some of the units not covered under the purview of SSI sector from the EC data file, by following a broad approach. The first obstacle in this elimination process was that no information was collected in the Economic Census on ‘value of plant & machinery’ and ‘value of fixed assets other than land and building’. Since the definition of SSIs and SSSBEs were based on these values, it has not been possible to eliminate those units, which did not conform to the definitions of SSI and SSSBE, in terms of these values. However, in order to get a usable data-file for preparing the sampling design for unregistered SSI sector, the following steps were taken.

4.7

The data-file extracted from the data of EC 1998 on the above lines was considered. It was found that the total number of units relevant for SSI sector was 79,04,146.

SAMPLINGS

4.8

The first task was determining the sample size, which will give the desired level of precision. The sample size at all India level was determined using the following formula.

d2 = t 2 (CV) 2 [1/n - 1/N], where
N is the size of the sector under study (assumed to be 79 lakhs),
CV is the coefficient of variation (assumed to be 7 based on Second Census),
t is the value of standard normal variate (= 1.96 for 95% confidence)
d is the permissible margin of error and
n is the sample size.

4.9

For d = 3 % and t = 1.96, n = 2,03,759. If simple random sampling technique is to be followed, a sample size of 2,03,759 at all India level would be expected to estimate the population parameters with a permissible margin of error of 3 % with 95 % probability. Keeping in view, the element of non-response, the sample size for the survey on unregistered SSI sector at all India level was fixed at 2,16,000.

4.10

Since the frame of unregistered SSI units was not available, a two-stage stratified sampling design was found suitable for the survey. The EC 98 data-file was used to identify homogeneous classes for the purpose of stratification. The first stage units (FSUs) were the census villages in rural sector and Urban Frame Survey (UFS) blocks carved out by the National Sample Survey Organisation (NSSO) in urban sector. The second stage units (SSUs) were enterprises falling in the unregistered SSI sector.

4.11

The EC - 98 frame was used for preparing the size-class distribution at State level separately for rural and urban sectors. The size was the number of enterprises and the size classes were 0, 1 to 10, 11 to 50, 51 to 200 and 201 & above. The total sample size of 2,16,000 was allocated to the size classes. The FSUs falling in the size class 201 & above, being out-liers, were earmarked for complete enumeration and the sample size in each FSU (i.e., no. of SSUs) in this size class was fixed at 20. After allocating the sample size to the size class 201 & above on these lines, the number of FSUs for the remaining size classes were fixed in proportion to the no. of enterprises, such that a minimum of 2 FSUs were allocated to each size class and the sample size in each FSU was 10 SSUs. For the purpose of this allocation, it was assumed that there would be one enterprise in each FSU in the size class ‘0’. The FSUs in each size class were arranged in the descending order of no. of enterprises and the required no. of FSUs for each size class were selected circular systematically with equal probability without replacement. By following this procedure, a total of 19,821 FSUs (10308 villages and 9513 urban blocks) were earmarked for the sample survey of unregistered SSI sector.

COVERAGE

4.12

After removing the duplicate records, it was found that a total of 19,579 villages/ urban blocks were surveyed by the State Directorates of Industries, out of which data for both FSUs and SSUs were received in respect of 19,278 only. For the remaining 301-villages/ urban blocks, data for FSUs were received but the data for SSUs was not received. Hence, the information of only 19,278 FSUs was used in preparing the estimates. In these 19,278 FSUs, the enumerators selected 1,68,654 unregistered enterprises for survey. However, they could finally survey 1,67,665 enterprises (99 %) and the remaining could not be surveyed due to non-response. The distribution of FSUs and SSUs for these 19,278 FSUs in each State is given in the following table.

TABLE 12: COVERAGE IN THE SAMPLE SURVEY OF UNREGISTERED SSI SECTOR

HOW UNREGISTERED SSI UNITS WERE IDENTIFIED AND SURVEYED

4.13

In each of the FSU selected for survey, the enumerators listed the units falling in unregistered SSI sector by following an elimination process. Initially, they identified the non-agricultural enterprises operating within fixed premises. Then they sought further information to identify whether the unit has been pursuing any economic activity relevant to SSI sector. At this stage, the units pursuing economic activities irrelevant to SSI sector were eliminated. The next stage followed was to find out whether the unit has been pursuing any economic activity, which is under the purview of Small Industries Development Organisation (SIDO), viz., of an SSI or a SSSBE. At this stage, the non-SIDO units and the Government/ pubic sector units were eliminated. From the remaining units, information was obtained to further classify the units in terms of upper ceiling limits prescribed (Rs. one crore original investment in Plant & Machinery for SSIs and Rs. 10 lakhs investment in fixed assets other than land and building for SSSBEs). The units falling above these ceiling limits were eliminated at this stage. The remaining units were eligible for registration as an SSI or SSSBE depending upon the economic activity being pursued by it. In the next stage, the enumerators sought details whether the unit was permanently registered with the respective District Industries Centre (DIC) as an SSI or SSSBE, as the case may be. In this stage, the registered SSIs and SSSBEs were eliminated. For the remaining enterprises, which were unregistered units of the SSI sector, the enumerators collected information on name and address and employment. These enterprises were grouped into two sub-strata, viz., SSIs and SSSBEs. The allocation of total sample size (i.e., 10 for FSUs falling in strata 1 to 4 and 20 for FSUs falling in stratum 5) was done by the enumerators proportionately to the size of the two sub-strata, subject to a minimum of 2 in each sub-stratum. The sample enterprises (SSUs) to be surveyed were selected by the enumerators using circular systematic sampling after arranging the units in each sub-strata in the descending order of employment.

4.14

A working sheet (Format-IV given in the Appendix) was prescribed for use by the enumerators for following all the above stages. The working sheet was not taken up for data processing. Instead, another format, viz., Format-II (given in the Appendix) was prescribed to be filled by the enumerator for each FSU to summarise the details in the working sheet. The detailed inquiry on selected enterprises (SSUs) was done in Format-III (given in the Appendix). The data received in Formats - 2 & 3 were processed using ICR technology.

4.15

Although, the approach was to begin the exercise by identifying the non-agricultural enterprises and end with the final selection of unregistered SSI sector units, some of the enumerators did not systematically follow this approach, as the data indicates. In about 3,083 villages/ urban blocks, the enumerators directly identified the unregistered SSI sector units. This has posed problems in using the information collected at various stages of the above elimination process for estimating the parameters other than that of unregistered SSI sector. In many cases, the classification of sub-strata was wrongly done as some of the units classified at the listing stage as SSIs were found to be SSSBEs during detailed inquiry and vice versa. In view of these limitations, only the SSU level data was used to estimate all the parameters.

ESTIMATION PROCEDURE

Let

‘p’ stand for sector (p = 1to 2)
‘q’ stand for State (q = 1 to 32)
‘r’ stand for stratum no. (r = 1 to 5)
‘s’ stand for sample serial no. (FSU No.) within a given p, q, r.
‘t’ stand for sub-stratum no. within a given p, q, r, s (t = 1, 2).
‘u’ stand for enterprise no. within a given p, q, r, s, t.

Let

‘N’ stand for the number in the population with the above suffixes.
‘n’ stand for the number surveyed in the sample with the above suffixes.
‘X’ stand for the total of a variable/ characteristic in the sample.

Npqr = No. of villages/ Urban blocks in the population for a given p, q, r. (These values were taken from the data-file extracted from Economic Census 1998.)

Let

n_pqr= No. of villages/ Urban blocks surveyed in the sample in a given p, q, r.

4.16

The values Npqr were further adjusted to cover non-response. Where no FSU could be surveyed within a stratum, the no. of FSUs in the Stratum were merged in the adjacent stratum to get estimates without deliberate underestimation. In case of Daman & Diu, the data on 6 rural FSUs was received but the corresponding data for SSUs was not received. Hence, the rural strata of Daman & Diu were merged with those of Dadra & Nagar Haveli and a combined estimate is provided for both these UTs in the results. After making these adjustments, the Multiplier at FSU level, i.e., for a given p, q, r, s is computed as (Npqr / npqr ).

Let

X_pqrstu = Value of any variable/ characteristic for a given p, q, r, s, t, u.

Estimate of the variable total for the population at all India level

4.17

(N_pqr / n_pqr) * (N_pqrst / n_pqrst) is the multiplier at SSU level. Before calculating this multiplier, adjustment was made in N_pqrst to cover total non-response at Sub-stratum stage. Where not even a single selected unit in a Sub-stratum could be surveyed, the corresponding size of the Sub-stratum was added to the size of the other Sub-stratum. The multiplier so calculated was appended to the datafile of SSUs at unit level as an additional item.

4.18

The stratification was done on the basis of Economic Census data, which contained the old structure of State and district codes, i.e., of 557 districts and 32 States/ UTs. However, for the selected villages/ urban blocks, the enumerators provided the new State code and district code in respect of 593 districts and 35 States/ UTs, as used by the Office of Registrar General of India in 2001 population census. These new codes along with the multiplier provided at unit level made it possible to generate estimates for the new States, after applying multipliers. However, in case of Delhi, the new district codes could not be fully followed, as the data collection mechanism failed in many cases to provide the correct codes. For this reason, most of the records pertaining to Delhi State bear the district code ‘01’ as a default option.

4.19

At the record level, it was found that in many records important information such as market value of fixed assets, original value of plant & machinery, gross output and exports were either not reported or underreported, as in the case of registered SSI sector. This could be possible due to non-response. In order to partially mitigate the extent of underestimation, market value of fixed assets and gross output being applicable to all types of units were estimated for the missing records using the average for the non-missing ones and the total estimates were prepared. In respect of original value of plant & machinery, estimate was prepared for the missing records belonging to SSI category using the average for the non-missing ones of the same category and the total estimate for SSI category was arrived at. For the SSSBE category, no estimation was done for missing observations, as plant & machinery may not be applicable to most of the units falling in this category. Similarly, for exports also, no estimation for missing observations was resorted to.