Non-farm Income and Consumption Expenditures in Rural Bangladesh: Empirical Evidence from Multilevel Regression Modelling

This study examines the effects of non-farm income on household consumption expenditures in rural Bangladesh. A two-stage endogenous treatment effect model is built on data from a nationally representative Household Income Expenditure Survey (HIES) 2010 to control selection bias. The HIES follows a hierarchical data structure because the survey is based on two-stage stratified sampling. A multilevel mixed-effects linear regression model is used to capture the unobserved heterogeneity between clusters (PSUs) along with revealing important factors. Results reveal that non-farm income has a significant positive effect on household’s consumption expenditures and non-farm income recipient households spend about 29% more than their counterparts. In addition, higher level of per capita income, education, smaller family size and lower dependency ratio are found to be more effective in increasing consumption expenditures of rural households. Significant cluster-level variations are observed in the analyses. This study recommends that non-farm income generating activities should be encouraged among rural households as this would raise their consumption expenditures and hence, improve welfare and living standards among them.

     

Assimilation of Time Series Data into a Dynamic Bioeconomic Fisheries Model: An Application to the North East Arctic Cod Stock

This paper combines the elegant technique of Data Assimilation and a Monte Carlo procedure to analyze time series data for the North East Arctic Cod stock (NEACs). A simple nonlinear dynamic resource model is calibrated to time series data using the variational adjoint parameter estimation method and the Monte Carlo technique. By exploring the efficient features of the variational adjoint technique coupled with the Monte Carlo method, optimal or best parameter estimates with their error statistics are obtained. Thereafter, the weak constraint formulation resulting in a stochastic ordinary differential equation (SODE) is used to find an improved estimate of the dynamical variable, i.e. the stock. Empirical results show that the average fishing mortality imposed on the NEACs is about 16% more than the intrinsic growth rate of the biological species.