Agricultural Research Management Information System

Home
Research Summary
Search
About Us
- ARMIS
- Brochure
Contact Us
Report
User Request
Data Input
Help
- Operation Manual
  - PDF
  - Video
- Program Area & Commodity
We have reached 37600 number of research entries at this moment.

- Logout

Research Detail

Home
Research
Detail

Targeting Vulnerable Small Farm Households in Bangladesh

Alastair Orr
Farming Systems Economist
Natural Resources Institute, UK

Noel Magor
Research Fellow
University of Adelaide

A.S.M. Nazrul Islam
Senior Scientific Officer
Bangladesh Rice Research Institute (BRRI)

Abstract/ Executive Summary:

A major challenge facing agricultural researchers and policy makers in Bangladesh is to improve the livelihoods of small farmers. Farms between 0.2-1.0 ha^-1 comprised 61% of total farm households in 1983/84 and will comprise 72% of total farms by the first decade of the next century (Ahmed 1987). One in five small farm households lives below the absolute poverty line while two in five have average incomes only 25% above the poverty line (BBS 1991). Low productivity and frequent natural hazards have made large numbers of these farms vulnerable to impoverishment. Although previous studies have analysed downward mobility among small farm households (van Schendel 1981; Rahman 1986), there is no clear definition of small farmer vulnerability on which policy-makers and researchers can base programmes or design appropriate ways to raise farm productivity. The general objective of this article is to define the target group of small farm households which are vulnerable to impoverishment through distress sale of land.The analytical approach follows that of Farming Systems Research and Extension (FSRE), which has a large literature on the identification of farmer target groups (Doorman 1984; Jamtgaard 1989; Jolly 1988; Manyong et al 1988). Specifically, the article attempts to: (1) identify the target group; (2) provide useful indicators for this target group; and (3) determine in which rice environments technology has the greatest potential to reduce small farmer vulnerability.

Keywords: Vulnerable,Small farm households

Location:

Start Date:

End Date:

Program Area: Socio-economic and Policy

Non-commodity: Social status

Objectives:

To define the target group of small farm households which are vulnerable to impoverishment through distress sale of land.

Methodology:

Data

The data derives from a study of small farmer vulnerability conducted by the Bangladesh Rice Research Institute (BRRI 1992). Data were collected from five major rice environments, where several villages were selected, and households stratified by land ownership were randomly sampled. Data on food security, land holding, land transfers, and credit arrangements were collected using structured questionnaires in three survey rounds between 1989-92.

Selection of Variables

Since cluster analysis cannot distinguish between relevant and irrelevant variables, meaningful results depend on selection of appropriate variables. In this analysis, variables to stratify rural households were selected using wealth-ranking, whereby key informants ranked neighbouring households according to income based on criteria used by villagers themselves (Grandin 1988; IIED 1992). Wealth-ranking has been shown to give results closely comparable to those obtained from conventional quantitative methods, such as income-expenditure surveys (Lanjouw and Stern 1991). Wealth-ranking excercises were conducted in three villages representing two rice environments (Orr et al forthcoming).

Data Preparation

Clustering data or objects requires some way of measuring the distance between them. This analysis used the common measure of the sum of the squared differences over all the variables (i.e., squared Euclidean distance). Since the measure of Euclidean distance depends on the units of measurement for the variables, all the variables used in the cluster analysis were first standardized to Z scores with a mean of 0 and standard deviation of 1. Outliers were identified by performing a preliminary cluster analysis and removing cases which formed clusters with only one observation. The final sample size was 1,162 households.

Cluster Analysis

Several methods exist to combine objects into clusters, with the objective being to maximise the differences between clusters relative to the variation within the clusters. Hierarchical methods begin by classifying each object as a separate cluster, gradually combining object into one large cluster. Non-hierarchical methods begin by selecting a cluster centre or 'seed points' (averages of the cluster variables) and all objects within a pre-specified distance are included in more accurate, non-hierarchical methods are widely used with large data-sets since they are computationally more efficient (Hair et al 1992). The present analysis used the non-hierarchical QUICK CLUSTER routine, which follows the parallel threshold method whereby seed points are randomly selected for a pre-specified number of clusters (SPSS Inc. 1990). The appropriate number of clusters was determined empirically by comparing the results of analyses with five to eight clusters. The cluster analysis was made in two stages. First, a preliminary cluster analysis was made to identify a sub-sample of 'poor' households using the variables generated through wealth-ranking. This analysis (not reported here) produced five clusters from the complete sample of 1162 households. Two clusters representing rich households and households with salaried employment were removed, leaving a sub-sample of 937 'poor" households. A second cluster analysis was then made to identify vulnerable farm households among these 937 'poor' households. Once again the five cluster solution was selected, with a total sample size of 382 households.

Source of Information: The Bangladesh Development Studies,Vol. 23, No. 1/2 (March-June 1995), pp. 29-47 Published by: Bangladesh Institute of Development Studies

Article Link (Online Journal):

Funding Source:

Total Budget (Tk.) :

Achievement/Findings:

An alternative approach is greater investment in rice research for marginal environments. Although the benefits from MV rice in Bangladesh have been evenly distributed between small and large farmers (Hossain 1989), MVs have yet to be developed for these marginal rice environments. This has been attributed to a flawed research strategy imposed by foreign donors which overlooked the diversity of Asian rice environments (Anderson et. al 1990). The formidable technical problems of developing MVs for marginal rice environments (Catling 1992), however, together with poor infrastructure, mean that this strategy to reduce small farmer vulnerability must be regarded as long-term. In the short to medium term, greater investment in downstream research to broaden the range of crop and non-crop interventions offer potentially high returns. Although FSRE in marginal environments is unlikely to produce technological breakthroughs, experience elsewhere suggests that the cumulative effect of small, incremental changes can be significant (Byerlee and Husain 1993).Finally, the disproportionate number of vulnerable farmers in the DWR environment suggests that flood-control might effectively reduce small farmer vulnerability. Closer analysis, however, revealed a much lower volume of distress sale of land at the Mirzapur study site, where more farmers had access to bank credit, or grew MV Boro after DWR (Orr et al 1992). Infrastructure development and irrigation to increase the resilience of potentially vulnerable small farmers, might be more effective and less costly than flood-control.

Knowledge Management: Journal