Bangladesh Census of Manufacturing Industries 1995-96
The Bangladesh Bureau of Statistics (BBS) has been conducting a Census of Manufacturing Industries (CMI) for the last two decades under the Industrial Statistics Act, 1942. It is designed to cover all industrial units regulated by the Factories Act, 1934 and registered with the Chief Inspector of Factories (CIF). The factories Act covers all units that employ 10 or more workers. These units may or may not use power in their industrial operations.
Before the CMI of 1989-90 all the previous censuses conducted were based only on those manufacturing establishments which were registered with the CIF under the factories Act 1934. The list of such registered manufacturing establishments was collected from the Chief Inspector of Factories and was used as a frame for conducting the previous censuses. The unregistered manufacturing establishments thus remained outside the scope of the previous censuses . In 1986 a Census of Non-Farm Economic Activities and disabled persons was conducted and the report thereof was published in 1989. Similarly, a census of Handloom Industries was conducted in 1990. The report of the census was published in January 1991. These two census-work provided the basis for covering the unregistered manufacturing establishments in the subsequent CMIs. Accordingly, in the subsequent CMI operations beginning from the 1988-89, steps have been taken to cover both registered and unregistered manufacturing establishments in the CMI. This has made the CMI more representative of the entire manufacturing sector. This CMI of 1995-96 provides basic statistics on the industrial structure and production of the manufacturing sector of the country on a more comprehensive basis.
As with the earlier CMIs the main objectives of the 1995-96 CMI were to determine the volume of industrial production, fixed capital formation, employment and gross value added in manufacturing industries. The data collected followed broadly the scope of the earlier CMIs, specifically in the following areas:
* ownership status;
* fixed assets by category;
* employment by category;
* value and quantity of raw material consumed;
* employment cost;
* operating expenses;
* inventories of physical assets;
* cost and quantity of fuel consumed and by category;
* value of gross output;
* value added and indirect taxes, etc.
The information collected was ultimately processed and classified by manufacturing sector activities following the Bangladesh Standard Industrial Classification (BSIC) at 4 - digit level.
Kind of Data
Census/enumeration data [cen]
Two independent list frames one for the handlooin establishments and the other for the manufacturing establishments(excluding handloom establishments)were used for the collection of data. The sample of handloom establishments was taken from an updated list of area frame of handloom establishments. The sample of manufacturing establishments(excluding handloom establishments)was taken from a directory of manufacturing establishments, compiled from the economic census of 1986. For the purpose of choosing the sample manufacturing establishments (other than handloom establishments) were classified as large and medium establishments on the basis of the number of persons engaged. A total number of 11,474 manufacturing establishments were taken as the sample for this census of which 2250 were handloom establishments.
Filled-in-returns from 4,853 establishments were collected from the sample frame of 11,474 establishments.
Producers and sponsors
Authoring entity/Primary investigators
BANGLADESH BUREAU OF STATISTICS
Statistics Division, Ministry of Planning
It was decided to cover all large scale manufacturing industries identified in the directory (excluding handloom) of the 1995-96 CMI. It was also decided that about 2,860 establishments be selected from the medium scale establishments.
The total sample size was determined on the basis of precision requirements and also for obtaining reliable results for as many 4 - digit level BSIC groups as possible. A stratified single-stage systematic sample design was used to select the sample of manufacturing industries. To increase the efficiency of the sample design the frame was stratified into as many homogeneous strata or groups as possible. After determining the approximate effect of a1ternative sampling strategies on the precision of the estimates at the 4 - digit BSIC level, it was decided to divide the manufacturing sector, represented by the directory and list frame of handloom establishments, into the following 7 substrata:
1. BSIC with 10-19 establishments.
2. BSIC with 20-49 establishments.
3. BSIC with 50-99 establishments and with less than 50% of overall TPE in the 50+TPE stratum..
4. BSIC with 50-99 establishments and with more than 50% of overall TPE in the 50+ TPE stratum.
5. BSIC with 100-299 establishments and with less than 50% of overall TPE in the 50+TPE Stratum.
6. BSIC with 100-299 establishments and with more than 50% of overall TPE in the 50+TPE $tratum.
7. BSIC with 300+establjshménts.
Since many of the characteristics of the manufacturing units covered by the CMI such as value of production, intermediate consumption, etc. are correlated with TPE (total persons engaged) the sampling frame was also stratified by TPE size groups.
The new sampling approach for the manufacturing sector resulted in a sample of approximately 9,224 manufacturing establishments (excluding handloom) which was within the upper limit dtermined by the resources of the CMI.
The handloom establishments were selected through an area sample. A two stage sample design was used for the purpose. The first stage was for selection of the enumeration areas which were selected with probability proportional to size (PPS). In the second stage, the ultimate sampling units or handloom establishments were selected by using systematic sampling method.
Deviations from the Sample Design
According to the specified sampling procedures, the basic weight for the establishments in each stratum would be equal to the sampling interval, and the certainty establishments would have a basic weight of 1. Following the data collection, these basic weights need to be adjusted for non-response. The adjustment procedures are explained in the following articles below.
The 1995-96 CMI data were collected by using list and area frames of establishments employing more than 10 persons. It was necessaiy for adjustment of basic sampling weights (expansion factors) to take into account of non-interview/non-response. Given the different nature of frames, the non-response/ non-interview adjustment procedures were different for the establishments in the two survey frames.
Adjustments for List Frame/Directory of establishments
In case of list frame for manufacturing establishments (excluding handloom), the basic sampling weights varied by stratum. The certainty stratum had a basic weight of 1. In order to ensure an appropriate non-interview adjustment to the weights, an interview status code was introduced for each sample establishment. The following 2-digist interview status codes were introduced to ensure proper adjustment to weights:
1. Completed questionnaire
2. Incomplete information-not used
4. Respondent not available
5. Mail-out questionnaire-not returned/not follow-up
6. Not found (after follow-up)
7. Transferred-no information
9. Closed permanent
10. Change of industry-out of scope
11. Number of TPE-out of scope
12. Fake establishment
14. New establishment-not in frame
Some of these codes applied mostly to the manufacturing sector, where the data collection began with a mail-out of the questionnaires, followed by personal interviews for most of the establishments which did not respond in time. The last category (14) was for new establishments not in the frame but which were identified and interviewed by the enumerators.
After the status code had been entered for each establishment in the file, the total number of sample establishments by status code for each stratum was calculated. The following noninterview adjustment factor was then calculated separately for each stratum:
Ah= No. of establishments with status codes 01 through 08 in stratum h
No. of establishments with status code 01 in stratum h
where Ah=non-interview adjustment factor for stratum h.
The rumerator of the adjustment factor was the number of valid establishments and. the denominator was the number of respondent establishments.
The adjustment factor was not applied to the weight of questionnaire with status code 14 (new establishments not in frame), since they were simply to be added to the file v'ith a weight of 1. The findd weight for the remaining establishments in stratum h were calculated by using the
where wh = basic weight for stratum h.
Adjutments for List/area frame for haiidloom establishments
For the area/list frame, each sample enumeration area had a unique weight, since the sample areas Were selected with probability proportional to sizc (PPS) and its weight was adjusted in the first stage of sampling. For this purpose, it was necessary to necessary to determine the number of valid (10+) establishments in the sample area and the number of completed interviews for each sample area. The basic weight for the sample area was multiplied by the following non-interview adjustment factor.
No. of valid establishments in the sample area
Ahi = ...................................................................
No. of respondent establishments interviewed in the area
where Ahi = adjustment factor for the i-th sample area in stratum h.
Dates of Data Collection (YYYY/MM/DD)
Mode of data collection
Type of Research Instrument
A questionnaire with an instruction manual in Bengali was used. The questionnaire was canvassed both for large and medium scale establishments as well as for handloom establishments. The CMI questionnaire is reproduced in the appendices.
Ministry of Planning
The filled-in returns were first edited manually. Omissions, inconsistencies and errors detected in the returns were corrected by the desk officers. In some cases respondents had to be contacted for elucidating further information. In a few cases where information gaps could not be; rectified, imputations were made. Coding of specific items was done manually. After editing and coding, the data entry work waws done be the Industry and Labour Wing in the MicroComputer environment. Final tables obtained from the computer in the form of computer printouts were directly used for publication. This helped avoiding corrections and editing at the printing stage and expedited considerably the printing process. For tabulation pulposes classification of the manufacturing establishments followed the Bangladesh Standard Industrial Classification (BSIC '86) which is developed in conformity with the International Standard Industrial Classification.
BANGLADESH BUREAU OF STATISTICS
Disclaimer and copyrights
The user of the data acknowledges that the original collector of the data, the authorized distributor of the data, and the relevant funding agency bear no responsibility for use of the data or for interpretations or inferences based upon such uses.