Testing for Complementarity: Glyphosate Tolerant Soybeans and Conservation Tillage

Campus Units

Economics, Center for Agricultural and Rural Development

Document Type


Publication Version

Submitted Manuscript

Publication Date


Journal or Book Title

American Journal of Agricultural Economics





First Page or Article ID Number


Last Page





Many decisions in agriculture are made over combinations of inputs and/or practices that may form a technology system linked through complementarity. The presence of complementarity among producer decisions can have far-reaching implications for market outcomes and for the effectiveness of policies intended to influence them. Identifying complementarity relations, however, is made difficult by the presence of unobserved heterogeneity. Drawing on recent methodological advances, in this paper we develop a test for complementarity between glyphosate tolerant soybeans and conservation tillage that overcomes certain limitations of previous studies. Specifically, we develop a structural discrete choice framework of joint soybean-tillage adoption that explicitly models both complementarity and the correlation induced by unobserved heterogeneity. The model is estimated with a large unbalanced panel of farm-level choices spanning the 1998–2011 period. We find that glyphosate tolerant soybeans and conservation tillage are complementary practices. In addition, our estimation shows that farm operation scale promotes the adoption of both conservation tillage and glyphosate tolerant seed, and that all of higher fuel prices, more droughty conditions, and soil erodibility increase use of conservation tillage. We apply our results to simulate annual adoption rates for both conservation tillage and no-tillage in a scenario without glyphosate tolerant soybeans available as a choice. We find that the adoption of conservation tillage and no-tillage have been about 10% and 20% higher, respectively, due to the advent of glyphosate tolerant soybeans.

JEL Classification

C35, D22, Q12, Q55


This is a working paper of an article from American Journal of Agricultural Economics 98 (2016): 765, doi: 10.1093/ajae/aaw001.