This study employs the multinomial endogenous treatment effect model to examine the effect of flood adaptation strategies on farm households' food security in the Upper East region, Ghana. In addition, an ordered probit model was used to analyse the determinants of household's recovery from flood shocks. Farmers adopt on-farm and non-farm activities as adaptation strategies. Estimation results indicate that farmers that employ on-farm and non-farm strategies had their food security situation improved and recovered faster from flood shocks. Age, education, access to extension, credit, farm size and information on flood occurrence drive the farmer's decision to adopt on-farm practices. Marital status, education, farm size and information on flood occurrence significantly influenced adaptation decisions related to non-farm activities. Other factors that influence household's recovery period from flood events were age, education, FBO and perceived severity of flood. Programs and policies that promote extension contacts, increase awareness on flood occurrences and provide non-farm work opportunities can be beneficial to reduce the adverse effects of floods.

Purpose: The purpose of this paper is to examine the factors that encourage households’ source separation behaviour in Accra and Tamale Metropolises in Ghana.Design/methodology/approach: Using a cross-sectional design, 855 households of Ghana were interviewed based on the theoretical framework of the theory of planned behaviour (TPB). The ordered probit regression model was employed to examine the factors that influence households’ source separation intention.Findings: The results indicated that educational attainment of head of household, total income of household, occupation type of household head, information, past experience with source separation, inconvenience in terms of time, space and availability of formal source separation scheme, attitude, subjective norm and the location of the respondents significantly predicted households’ solid waste separation intentions.Research limitations/implications: The cross-sectional design does not determine causality but an association. Thus, future studies should examine actual household waste separation behaviour by using the experimental design to test the TPB model.Practical implications: To promote solid waste separation at source, the public should be educated and provided with solid waste separation schemes that are efficient and compatible with households’ preference.Originality/value: This study was partly motivated by the fact that despite the benefits associated with source separation, little attention has been given to formal source separation in Ghana. Moreover, there are limited studies on source separation behaviour in Ghana using the TPB as the theoretical framework.