Canonical Correlation Analysis of bidirectional Linkages Between Agricultural Productivity and Household Welfare Development in Nepal

Abstract

This study explores how agriculture and household welfare are connected in Nepal using census data from 2001, 2011, and 2021. Two sets of indicators were created to represent structural changes in agriculture and household living conditions. All variables were standardized, and data checks using KMO, Bartlett’s test, and Cronbach’s alpha confirmed that the indicators were suitable for multivariate analysis (KMO: 0.76–0.84; Bartlett’s p-value<0.01; alpha: 0.75–0.89). Canonical Correlation Analysis was then applied to measure how the two domains move together, showing strong correlations across all years (0.90–0.94, p < 0.01). Despite the canonical R² for the first canonical variate being 81–86%, redundancy analysis revealed that the agricultural variate accounted for approximately 16–17% of the variation in the household welfare variate, whereas the household welfare variate explained 11–12% of the variation in the agricultural variate, indicating an asymmetry between agriculture and household welfare. The results suggest that while agriculture and household welfare are strongly linked, improvements in agricultural structure contribute more to household welfare than the reverse. This research provides an empirical framework demonstrating Nepal’s gradual shift from a fragmented agricultural economy to a more connected, although uneven, growth economy. The results emphasize the critical need for integrated development policies that address both agricultural transformation and persistent socio-economic disparities to foster equitable and
sustainable development across the nation.