At the farm level, this implies shortages of rice seeds, irrigation water and chemical inputs.

To analyse SRI adoption and impacts, we conducted a survey of close to 400 randomly selected households and also collected soil samples from farmers’ rice plots. The data show that SRI adoption rates vary regionally, and that even among the SRI users partial adoption is commonplace. Partial adoption implies that farmers use SRI techniques only on part of their total rice area. In addition, different SRI components are adopted to varying degrees. Regression analysis demonstrates that socioeconomic household characteristics tend to influence SRI adoption decisions. Participation in special training programmes is an important factor, because SRI is a knowledge-intensive technology. Moreover, sufficient availability of family labour is an important determinant of adoption. SRI is relatively labour-intensive, and, especially in the early phase of adoption when farmers are experimenting with the new system, family labour cannot easily be replaced by hired labour.

However, socioeconomic characteristics alone can explain adoption only to a limited extent.