A novel approach to dynamic livelihood clustering: Empirical evidence from Nepal

Rural households are heterogeneous: different socio-economic characteristics and asset endowments dictate their engagement with different livelihood activities resulting in different livelihood outcomes. Poverty reduction policies should consider this. Using a unique environmentally augmented three-wave panel dataset from 427 households in three locations of Nepal, we proposed an approach that combines households’ income and assets to identify different livelihood strategy clusters. Based on a Latent Markov Model we identify seven distinct livelihood strategies and analyse households’ movements between strategies through time. Most sampled households changed their livelihood strategy at least once between 2006 and 2012, and very few households transited directly from the least to the most remunerative strategy. A common pathway out of poverty appears to have involved an intermediate step during which households accumulated assets and capital through farming, petty trading and migratory work. The applied approach of combining income and assets better distinguishes the identified livelihood strategies compared to both the income and the asset approach and allows targeting of interventions towards specific strategies and transition pathways.