India - Trade Survey , January 1997 - December 1997, NSS 53rd Round

A stratified two-stage sampling design was adopted for the survey. The first stage units (FSUs) were the villages (panchayat wards in case of Kerala) in the rural sector and UFS blocks in the urban sector. The second-stage units were the OATEs and NDTEs for the trade survey.
The total all -India sample size of FSUs was allocated to different States/U.Ts in proportion to the number of workers engaged in OATEs/NDTEs in the respective States/U.Ts ( estimated as number of OATE/NDTE as per EC '90 multiplied by estimated number of workers per enterprise based on NSS 46th round) subject to a minimum allocation given to the State/U.Ts depending upon the total number of districts and town classes for the State/U.T. The State/U.T level sample size was again allocated to the rural and urban sectors in proportion to the number of workers.

Rural sector design

Sampling frame : The list of villages showing number of OATEs and NDTEs as per 1990 Economic Census was used for selection of villages in the States/U.Ts wherever such frame was available. For Kerala, list of Panchayat wards giving count of OATEs /NDTEs has been used as the frame for selecting Panchayat wards f or the the survey instead of villages. For Jammu & Kashmir, the
1981 census lists of villages forms the frame. For Andaman & Nicober islands, Lakshadweep and 5 districts of Madhya Pradesh, 1991 census lists were used as frame. For Andaman & Nicober Islands, villages remaining inaccessible throughout the year were excluded from the frame. For Nagaland, onlyvillages connected by bus or situated within 5 kms of the bus route were included in the frame.

Stratification : Each district generally formed a broad stratum. However, for Gujarat, where NSS regions cut across district boundaries, parts of each such part of a district formed a separate stratum. If any district (or part thereof lying in an NSS region in case of Gujarat) had a small number of trading enterprises, it was clubbed with a neighbouring district to form a broad stratum in order to ensure a minimum allocation. To net an adequate number of NDTEs in the sample, each broad stratum was divided into two area types : (i) Area type 1 consisting of villages having at least one NDTE, and (ii) Area type 2 consisting of the remaining villages of the broad stratum.

Where population census frames were used for selection of fsus , there was no division as above. In such cases, all the villages were classified in area type 2.

Allocation of sample villages among strata and area types : The State/U.T level rural sample size was allocated among the strata in proportion to number of workers. While allocating as above, it was maintained that NSS region level allocation were multiples of 8 and stratum level allocation is at least 4 but in multiples of 2. This was done in order to allocate at least two fsu's to each
of the area types. The stratum level allocation was again di stributed between two area types in proportion to number of NDTEs and OATEs taking into consideration that allocation for each area type was in multiples of 2.

Selection of fsu's : Villages were selected in the form of two independent sub-sampl es from each broad stratum X area type using circular systematic sampling with probability proportional to size, the size being the number of (OATE + NDTE) for area type 1 and the number of OATEs ( after assigning a size of 1 to the fsu's having no tradi ng enterprise) for area type 2.

Where population census frame was used, villages were selected using circular systematic sampling with probability proportional to population. For Lakshadweep and A & N Islands, however, equal probability sampling was adopted.

Urban sector design :

Sampling Frame : The latest available list of UFS blocks was used as sampling frame for selection of fsu's for all cities and towns.

Stratification : Town classes (broad strata) were formed within each district by grouping cities/towns according to population sizes as per '91 census. To net adequate number of trading enterprises in the sample, each town class was divided into two area types. Area type 1 consisted of the UFS blocks designated as 'bazar area' and area type 2 consisting of the remaining blocks of the town class.

Allocation of sample blocks among districts, town classes and area types : The state/u..t. level urban allocation is allocated among the districts and town classes in proportion of the number of workers. It is, however, ensured that NSS region level allocations were in multiples of 8 and town class level allocations were at least 4 and were in multiples of 2. Town class level allocations were further
allocated between two area types in such a way that UFS blocks of area type 1 get completely surveyed (central and state samples combined) subject to a maximum of 50% of allocation for town class level. Area type wise allocations were in multiples of 2.

Selection of blocks : For both the area types, sample blocks were selected circular systematically in the form of two independent sub-samples with equal probability.

Sampling of enterprises : All the OATEs and NDTEs of the selected village/block excluding those in the public sector and which operated for at least 30 days (15 days in case of seasonal enterprises) during the last 365 days preceding the date of survey were considered for sampling. All such, eligible enterprises were classified into three classes termed as 'enterprise class' according to
the total number of workers (including hired workers) employed in the enterprises on a regular basis.
The classes were as follows :


---------- ------------------------------------------------------------------------
Enterprise Composition of the class
class code
------------------------------------------ ----------------------------------------

1 all OATEs and NDTEs with one worker only.

2 all OATEs and NDTEs with two workers only.

3 all OATEs and NDTEs with three or more workers.
-----------------------------------------------------------------------------------

Before selecting enterprises from enterprise classes, all the enterprises within an enterprise class were arranged in ascending order of their NIC codes. The number of enterprises selected from enterprise classes 1, 2 and 3 were 4, 6 and 6 respectively. If there is a shortfall in enterprise class 3, it is made up from enterprise class 2, if possible, otherwise from enterprise class 1. Similarly, if there is a shortfall in enterprise class 2, it is made up from enterprise class 3, failing which from enterprise class 1. The priority order for compensation is 3rd, 2nd and 1st enterprise class.

The required number of enterprises were selected from each enterprise class circular systematically with equal probability.


There was no deviation from the original sampling design


The number of first-stage units (villages in the rural areas and UFS blocks in the urban areas) allotted were 6055 and 7169 but surveyed were 5988 and 7138 in rural and urban areas, respective­ly in the central sample. Under the State sample, 6530 villages and 8346 blocks were surveyed.


Three different weightsare provided in each record of filess in the data set. Deatils are as follows:-

1. Weight for each sub-round (Sub-round wise weight) is stored in Variable name : Wgt_ss

2. Weight for all Subrounds pooled and combined subsample weight is stored in Variable name : Wgt_combined