Asian Journal of Agricultural Extension, Economics & Sociology

In Elgeyo Marakwet County, Kenya's third-most potato-producing region, over 84% of smallholder transactions flow through broker intermediaries, not because farmers prefer this arrangement, but because institutional barriers systematically exclude them from higher-value governance structures. However, empirical studies on smallholder market participation across sub-Saharan Africa have two persistent limitations: most treat participation as binary, obscuring substantial heterogeneity across arrangement types, and almost none correct for the endogeneity of arrangement choice, because the transaction costs a farmer faces are themselves shaped by the arrangements in which they participate, rendering standard regression estimates directionally misleading. This study addresses the following research question: What institutional, socioeconomic, and structural factors determine smallholder potato farmers' participation in different clusters of institutional arrangements, after accounting for the endogeneity inherent in the choice of arrangements? Using cross-sectional data from 577 smallholder potato farmers sampled via multi-stage systematic random sampling, we apply a control-function multinomial logit (CF-MNL) model, instrumenting four endogenous transaction-cost and enforcement variables to address the simultaneity between observed transaction costs and arrangement choice, with ward-level historical contract prevalence and community-institution indices serving as instruments. First-stage F-statistics confirm instrument relevance and strength: transport costs (F = 167.30), information search costs (F = 17.08), negotiation costs (F = 10.55), and informal sanctions (F = 18.48). Endogeneity tests confirm severe bias in uncorrected estimates (χ² = 51.99 to 1,303.24, p < 0.001), and model fit improves from Pseudo-R² = 0.160 to 0.183 following correction (LR χ² = 27.54, p < 0.001). After correction, access to extension services emerges as a consistent positive determinant of higher-value arrangements. Farm size serves as a volume threshold, confining small-farm households to low-intensity broker arrangements regardless of motivation. Gender effects are large, persistent, and unexplained by observable endowments, consistent with the presence of gender-related barriers to institutional access that operate beyond the measured characteristics and that generic market development programmes may not adequately address. These findings are consistent with the interpretation that participation is not a matter of farmer preference but of institutional access. Drawing on comparative evidence from SSA countries, the study identifies extension reorientation, aggregation infrastructure, and gender-responsive instruments as the most actionable policy responses.