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

Determinants of Livelihood Choices: an Empirical Analysis from Rural Bangladesh

Sanzidur Rahman
Corresponding author and Associate Professor in Rural Development
School of Geography, Earth and Environmental Sciences, University of Plymouth, UK

Shaheen Akter
Freelance development consultant, Milton Keynes, UK

Abstract/ Executive Summary:

The study identifies socio-economic determinants of livelihood choices of rural households in Bangladesh (4,195 households from 139 villages) by applying a multinomial logit model of occupational choice and a multivariate Tobit model that allows for jointness in decision making. Results reveal that households choose multiple livelihood options. A number of socio-economic factors and resources at the household level as well as the state of rural infrastructure significantly determine households’ choice of livelihood options. Overall, resource-rich and educated households engage in diversified livelihoods and rural infrastructure promotes diversification of livelihoods. Female-headed households fail to participate in any of the livelihood categories and consequently earn significantly lower incomes. Policy implications include investment in rural infrastructure, irrigation, rural electrification, education, livestock resources, as well as a targeted approach for female-headed households, e.g., creation of a hired labour market and skills/education programs for females.

Keywords: Livelihood choices, Rural infrastructure, Multinomial logit model, Multivariate Tobit model, Bangladesh.

Location: In Bangladesh

Start Date:

End Date:

Program Area: Socio-economic and Policy

Commodity: Livelihood

Objectives:

The objective is to identify the factors that influence the choice of or between multiple livelihoods as well as the intensity of involvement as reflected by the income earned from each activity.

Methodology:

Modelling factors influencing livelihood choice: multinomial logit model and its limitation. A critical empirical question is how to define multiple livelihood options for the purpose of quantitative analysis. Although multinomial logit models have some advantages in examining livelihood diversity, there are two main concerns with letting dichotomous variables represent livelihood options. First, the zero-cutoff is problematic since a farming household will diversify income sources by choosing agricultural and non-agricultural options, simultaneously. The standard solutions are counts of a minimum number of days or minimum proportion of income as a cutoff. For example, Hatlebakk (2012) introduced a minimum number of days threshold before an occupation counts. Important information is lost by such arbitrary cut-offs. A significant income may also be wrongly categorized. For example, a household may draw 35% of income from agricultural sources and 65% of income from non-agricultural sources. An arbitrary cut-off of 40% would place this household in the non-agricultural group, resulting in substantive loss of information. To avoid this loss, we use income from different options as a measure of the intensity of choice. Second, the dichotomous dependent variable fails to consider the variation within the 0-1 range (choose an option or not). However, when an option is chosen it is possible to measure the intensity of participation by using a continuous variable above zero by income earned or number of days of work. This is why we use income as dependent variable and multivariate Tobit model for estimation. In what follows, we first estimate a multinomial logit model of occupational/livelihood choice (i.e., Eq. 1) for setting the scene and shedding light on the motivations underlying livelihoods choice. Also the model provides rich information on factors influencing various occupational choices with reference to a base category. We then proceed to estimate a multivariate Tobit (three equations) model of livelihood choice. Our conclusions are drawn from the results of both models. 2.2 Modelling factors influencing livelihood choice: multivariate Tobit model We postulate that the households follow sequential decisions; first ‘whether to participate in a particular livelihood option or not’; and second, conditional on participation, ‘what is the level or intensity of participation? In such a case, a censored regression model is required. A Tobit model is the most suitable because it uses all observations, both those at the limit, usually zero (e.g., non-participants), and those above the limit (e.g., participants), to estimate a regression line as opposed to other techniques that use observations which are only above the limit value (McDonald and Moffit, 1980). The procedure also captures latent level of intensity of potential households who decide not to participate in a particular livelihood option. 2.4 Variables used for empirical estimation of the models The empirical application focuses on the identification of socio-economic factors that affect household livelihood choice decisions. The key elements and implicit assumptions underlying the models are (based on Adbulai and CroleRees, 2001): (a) rural households have access to land and labour as production inputs. Labour can be either provided by the household or hired. (b) the market for land is either missing or imperfect; (c) all households are engaged in agriculture; (d) diversification to other activities (e.g., non-agricultural activities) require either capital or skills; and (e) households may diversify livelihood activities to spread income risks.

Source of Information: Journal of South Asian Development, 9(3): 287-308, December 2014

Article Link (Online Journal): https://pearl.plymouth.ac.uk

Funding Source:

Budget:

Total Budget (Tk.) :

Achievement/Findings:

This study identifies the determinants of livelihood choices and the income derived from such choices of the rural households in Bangladesh using multinomial logit and a multivariate Tobit model. While the multinomial logit analysis enables the identification of factors influencing the choice of livelihood options, the multivariate Tobit model allows joint determination of multiple livelihood choices and the income derived from such choices. The model diagnostics confirmed the jointness of the decision on livelihood choice, thereby, justifying our use of the multivariate approach. The results reveal that a number of socio-economic factors as well as the state of rural infrastructure affect households’ livelihood choice decisions. The multinomial choice model reveals that land-rich households are less likely to combine agriculture with non-agricultural livelihoods. Education influences non-agricultural livelihood while livestock resources influence agricultural livelihood. State of rural infrastructure influences diversification of livelihoods. Results from the multivariate Tobit model reveals that resource-rich and educated households with younger heads in villages with access to irrigation but underdeveloped infrastructure are more likely to choose agricultural livelihood. Proximity to regional headquarters also influences choice of agricultural livelihood. Wage employment is chosen by resource-poor and uneducated households in villages with no access to irrigation but with developed infrastructure and closer to the district or regional headquarters.

Knowledge Management: Journal