Improved seeds and rice productivity in Tanzania: policy insights from national sample census data

Improving rice productivity is central to Tanzania’s efforts to strengthen food security, reduce rural poverty, and advance agricultural transformation. Although improved seed technologies have been widely promoted, robust national-scale evidence on their performance under real-world smallholder conditions remains limited. This study aims to assess how improved seed adoption influences rice productivity across agroecological zones (AEZs) in Tanzania, addressing a key empirical gap in the existing National Sample Census of Agriculture (NSCA) literature. The study draws on nationally representative microdata from the 2019/20 National Sample Census of Agriculture (NSCA), complemented by the 2007/08 NSCA for normalization. A nonparametric stochastic simulation approach was applied to generate empirical yield probability distributions for farmers using local seeds and improved seeds across nine AEZs. Using Latin Hypercube Sampling, the analysis estimates the likelihood that rice farms exceed nationally relevant (3.0 t/ha) and globally benchmarked (4.5 t/ha) productivity thresholds. Unlike prior NSCA analyses relying mainly on descriptive or regression-based methods, this simulation-based framework explicitly captures yield variability and production risk. Results show that farmers using improved seeds consistently exhibit higher simulated yield probabilities than local-seed users. At the national level, improved-seed users are substantially more likely to exceed both the 3.0 t/ha national benchmark and the 4.5 t/ha high-yield threshold, while also facing a lower risk of sub-optimal yields. Productivity gains are most pronounced in high-potential zones, particularly the Southern Highlands and Eastern Zones, though improved seeds also reduce downside risk in more constrained agroecological settings. These patterns demonstrate that improved seed adoption shifts the entire yield distribution upward rather than merely increasing average yields. The findings provide national-level evidence that adopting improved rice seed varieties significantly enhances yield performance and reduces production risk across Tanzania’s agroecological zones. However, the magnitude of these benefits varies spatially, emphasizing the need for region-specific targeting. These insights can inform ongoing policy efforts under ASDP-II, TSSDS 2030, and NRDS-II to promote improved seed access and use, especially where potential gains are highest. Not applicable.