Analytical Report on Methodologies of Crop Estimation & Forecast and Private Stock of Food Grain Survey 2016-17
Agriculture sector plays a pivotal role in the economy of Bangladesh. The contribution of this sector to GDP is more than 15% and more than 43% of labour force are engaged in this sector. The performance of this sector has great impact on major macroeconomic objectives like employment generation, poverty alleviation and food security. Despite shrinking arable land due to overwhelming population pressure, the country has achieved remarkable progress in food production. Reliable and timely statistics on crop production and market situation are very essential for effective decision making on food management, export-import policy and appropriate planning of agricultural extension programes.
The Strengthening Agriculture Market Information System (AMIS) in Bangladesh project has been implemented by Bangladesh Bureau of Statistics (BBS) with the technical and financial support of FAO. The Department of Agricultural Extension (DAE), Department of Agriculture Marketing (DAM), Bangladesh Meteorological Department (BMD), Space Research and Remote Sensing Organization (SPARRSO) and Food Planning and Monitoring Unit (FPMU) were involved as coordinating agencies of this project.
The goals of the project were implemented in accordance with the strategic goals of National Strategy for the Development of Statistics (NSDS). The first objective of the project is to review and improve existing crop estimation system followed by BBS (Strategic Goal- 1.1). The second objective is to review and improve crop monitoring and production forecasting system of BBS (Strategic Goal -1.1 & 2.1). The third objective is to assess the situation of private stock of food grain (paddy, wheat & maize) in line with Strategic Goal - 4. The fourth objective of the project is capacity building of staff concerned with crop statistics (Strategic Goal -1 for Training Institute).
All these activities were implemented systematically and through extensive assessments to achieve the strategic goals of NSDS as mentioned. Reviewing and Improving of Existing Annual Crop Estimation Methodology The existing crop estimation methodology (area estimation) has been critically reviewed by project personnel (consultants and project implementation team). Especially, the cluster system that was developed in early eighties on the basis of Agriculture Census 1960 (where 10% mauza was covered) was critically reviewed. The time series area estimates obtained from cluster system were reviewed and found inconsistent as a whole. As the survey frame was not updated with any of successive censuses conducted in 1977, 1983, 1996 or 2008, the system was found ineffective. No documentation on sampling scheme is available for those clusters developed in eighties indeed.
The new methodology for area estimation of crop has been developed by AMIS project on the basis of list of mauzas with relevant crops derived from the latest agriculture census. Using the survey frame on the basis of the latest census is prime essence of any relevant surveys. In that
sense, the new methodology is statistically sound and updated. The main features of new methodology are (i) stratified two-stage sampling design where mauza has been considered as first-stage sampling unit (Primary Sampling Unit-PSU) and the producing farmers of relevant crops are the second stage sampling unit; (ii) list frame will be followed; (iii) the current year area will be derived using the ratio of aggregate cultivated area of current year by the related farmers in relation to previous year; (iv) twenty-five (25) mauzas will be selected according to probability proportion to sizes with at least one (1) mauza in a upazila and twenty (20) farmers to be interviewed from each mauza; (v) administrative advantage of the organizational set up of BBS has been considered. The new methodology has been discussed in national workshop held on 18 May 2016 and improved as per recommendation of the workshop.
The new methodology was reviewed in Technical Committee (headed by Director General, BBS) meeting on 1 January 2017 and was recommended for implementation. Upon the recommendations of the Technical Committee, the Steering Committee headed by Secretary, Statistics and Informatics Division, Ministry of Planning approved the methodology on 9 February 2017.
Use of Remote Sensing Technology in area estimation has also been recommended for the purpose of validation. The new methodology got technical approval from FAO accordingly.
Reviewing and Improving of Crop Forecasting Methodology The existing crop forecasting methodology has been reviewed. Similarly, the cluster system that was found obsolete and new methodology has been recommended. Ten (10) mauzas will be selected as sub-samples from the original twenty-five (25) sample mauzas of crop estimation. Twenty (20) farmers will be interviewed from each mauza. Data on cultivated area will be collected from these ten (10) mauzas at the initial stage of cultivation of relevant crop and data on tentative yield will be collected several days before crop is harvested. A rapid Assessment Report on crop forecast will be prepared between the two forecasts.
An econometric model has also been developed to provide an interim forecast of crop production using time series data. The use of remote sensing technology has been proposed for providing crop forecasts. All these were discussed in the international seminar on crop forecasts conducted by the project on 25-26 May, 2016. These methodologies were improved on the basis of the recommendation of the seminar. These were finalized after discussing the methodologies in relevant national workshop conducted by the project on 18 June 2016. These methodologies were approved by Technical Committee and Steering Committee along with the crop estimation methodology.
Kind of Data
Sample survey data [ssd]
Producers and sponsors
Authoring entity/Primary investigators
BANGLADESH BUREAU OF STATISTICS
STATISTICS AND INFORMATICS DIVISION, MINISTRY OF PLANNING
Statistics and Informatics Division
As stock of food grain is defined as the quantities of particular crop held in storage by supply chain agents from farmers to consumers at a given time, the farm and non-farm households and commercial agents are the sampling units of this survey. The non-farm users of food grains are everywhere but the farm households are not always found equally in all places for rice (aus, aman and boro), wheat and maize crops. Hence, an integrated method of area and list frame is recommended for the household level survey. On the other hand, survey at millers' level is more comprehensive and list frame is adopted. For millers' level survey, the exhaustive list has been used for selecting entities throughout the country on the basis of SRSWR. For household survey cultivating and non-cultivating households are the ultimate sampling units for rice survey. But only cultivating household are considered for wheat and maize survey as the non-cultivating households do not significantly buy or sale wheat and maize for consumption purposes. The cultivated mauza for specific crop has been considered as the Primary Sampling Unit (PSU).
Universe of Estimation
The whole geographic area of Bangladesh will constitute the universe of estimation. The estimates are required at national level. No district or division level estimation are required by the objective of the survey, only national estimates will be provided.
Considering the essence of estimates at national level and budgetary limitations, any stratification was not done at urban-rural or division/district level. So the data of any crops under this survey cannot analyse on this domain except national level.
Sampling Units The producing and non-producing farm households and millers are the sampling units of this survey. The non-farm users of food grains are everywhere but the farm households are not always found equally in all places for rice (Aus, Aman and Boro), wheat and maize crops. The cultivated mauzas for specific crop (Wheat and maize) have been considered as the Primary Sampling Unit. The exhaustive list of commercial units (millers) will be used as the frame for commercial agents' survey.
Stages of Sampling Two -stage stratified sampling design is adopted for the survey of each crop (rice, wheat and maize). The mauzas are treated as the first stage units (PSU) and the cultivating and non-cultivating household of the specific crop are the final or second stage units (USU). For survey of commercial farms/miller, the unit of farm/miller will be considered as the sampling unit. An exhaustive list of farms/mills will constitute the frame for such survey. Such selection will turn into one-stage selection phenomenon. Thus, sampling process will be more effective.
Sampling Frame An integrated method of area and list frame is used for the household level Stock Survey. On the other hand, survey at miller's level is more comprehensive and list frame is adopted. For miller's level survey, the exhaustive list has been used for selecting entities throughout the country on the basis of SRSWR. For household survey cultivating and non-cultivating households are the ultimate sampling units for rice survey. But only cultivating households are considered for wheat and maize survey as the non-cultivating households do not significantly buy or sale wheat or maize for consumption purposes. The cultivated mauza for specific crop has been considered the Primary Sampling Unit. So, the list of cultivating mauzas will constitute the first stage sampling unit for specific crops. The household in the mauzas have been considered as the second stage sampling unit. The commercial agent (miller) is directly selected from the list. In such case, the design will be turned to one-stage design.
Considering the possibility of getting required number of farmers in a district, a. district has been considered eligible which has area at least 10000 acres of cultivation of the crop under survey. Under this criterion, the list of district for aus aman, boro, wheat and maize crops were selected.
To address the coverage of the individual crop at national level, minimum sample household required for the survey has been calculated using the usual sample size determination formula;
Za 2 n = (CV(X)) xdeff. Where, CV(X) = is the coefficient of variation of the key variable za = is the value of the standard normal variate 2 d =is the proportion of allowable margin of error relative to the average crop area sizes deft = is the design effect. Usually, seen as 1.8 in most of the sample surveys of Bangladesh.
Taking d= .065 (6.5%) & CV (Land size) = 1.3 from Agricultural Census.
Sample Size for Rice:
1.962 n- .0652 (1.32) x 1.8
3.84 n-.004225 (1.32) x 1.8
n= 908.875x (1.69 ) x 1.8
n = 2765
However, we can take a sample of 3000 household for the rice survey. Taking 25 household from each mauza, we can select 120 mauzas as PSU for the survey.
Sample Size for Wheat:
Taking CV of Land size = 1.3 & d = .070 (7.0%) Za2 n— * (CV(X))2 x deff. n= 1.962.0049 (1.32) x 1.8 1 n = 3.84 — x 1.69 x 1.8 .0049
n = 783.67x 3.042 n = 2384
Taking 2500 households for wheat survey & Taking 25 households per PSU (2500/25 = 100) It has been selected 100 PSU for wheat survey.
Sample Size for Maize:
Taking CV of Land size = 1.3 & d = .070 (7.0%)
za2 n-22 (CV(X))2 x deff. r n= 1.962th (1.32) x 1.8 n = 3.84 1 x 1.69 x 1.8 .0049 n = 783.67 x 3.042 n = 2384
Taking 2500 households for Maize survey & Taking 25 households per PSU (2500/25 = 100) It has been selected 100 PSU for Maize survey.
Dates of Data Collection (YYYY/MM/DD)
Mode of data collection
STATISTICS AND INFORMATICS DIVISION
MINISTRY OF PLANNING GOVERNMENT OF THE PEOPLE’S REPUBLIC OF BANGLADESH
Data dissemination is similarly important as like as data collection and compilation. It is important to have a data dissemination calendar so that the stakeholders can be sure of getting data according to the calendar. Furthermore, immediately after the approval of data by the government, it should be uploaded in to the website for public use.
BANGLADESH BUREAU OF STATISTICS
BANGLADESH BUREAU OF STATISTICS
STATISTICS AND INFORMATICS DIVISION (SID) MINISTRY OF PLANNING GOVERNMENT OF THE PEOPLE’S REPUBLIC OF BANGLADESH