Astrid Mastenbroek

This thesis examines why adoption of improved maize varieties remains limited in Uganda despite well-documented yield benefits and decades of investments in breeding, dissemination, and extension. The objective of this thesis is to develop and empirically test a behaviorally grounded account of smallholder seed investment decisions under risk and uncertainty. Drawing on agricultural and development economics, behavioral economics, and seed-systems governance literature, the thesis analyzes how farmers interpret and respond to information and incentives within the institutional conditions of Uganda’s maize seed sector, including credence-good characteristics, and the gendered social roles and responsibilities.

To guide this analysis, the thesis draws on two complementary frameworks. Dual-process theory distinguishes between reflective decision-making, which involves deliberate evaluation of information, prices, and risks, and automatic processes, which rely on habit, affect, and experiential cues. The COM-B framework conceptualizes adoption behavior as the outcome of capability (knowledge and skills), opportunity (physical access and social context), and motivation (reflective and automatic), linking individual decision making to the institutional conditions of the maize seed sector and the social and cultural context in which seed decisions are made. The COM-B framework is primarily used as a diagnostic and organizing framework in the behavioral assessment (Chapter 2) and synthesis (Chapter 7). Throughout this thesis, adoption is operationalized as one-off demand, purchase or investment decisions rather than sustained multi-seasonal use.

Against this background, Chapter 1 situates the puzzle of low adoption within the adoption literature and argues that, while structural constraints and heterogeneity remain fundamental, they do not fully account for observed patterns. The chapter motivates closer attention to behavioral and psychological mechanisms and their interactions with economic constraints.

Empirically, the thesis adopts a mixed-methods approach that combines participatory qualitative research with incentive-compatible framed field experiments. Qualitative insights are used to diagnose behavioral constraints and gendered risk perceptions within the seed system. These insights inform the design and interpretation of experimental tests of valuation and risk salience mechanisms.

Chapter 2 develops a qualitative behavioral diagnosis of Uganda’s maize seed sector using the COM-B framework. Drawing on focus group discussions and triangulated with empirical studies from the maize seed sector in Uganda, the thesis shows that capability gaps are common but appear less decisive in the diagnostic than constraints related to opportunity and motivation. Low varietal turnover, thin retail networks, and lemon market dynamics driven by information asymmetry and weak enforcement of certification undermine confidence in formal markets. Automatic motivation and social opportunity shape seed choices through habits, trust in familiar seed sources, peer influence, and household and community norms, which filter what is seen as feasible and legitimate.

In Chapter 3, the role of information in shaping demand for certified maize seed is analyzed empirically. Using incentive-compatible auctions, the experiment tests whether providing certification and seed-quality information raises willingness to pay for a twokilogram seed pack. The findings show that while information improves knowledge of certification procedures and seed quality, it does not increase average willingness to pay. This suggests that information alone is unlikely to raise demand under low-credibility conditions.

In Chapter 4, reference-dependent preferences in farmers’ valuation of hybrid maize seed are examined. Using incentive-compatible auctions and a multiple price list, the experiment varies endowment status, price expectations, and ownership expectations for a one-kilogram pack of hybrid seed. The results show a large and robust endowment effect. Willingness to accept compensation for giving up seed substantially exceeds willingness to pay to acquire it. By contrast, experimentally induced price and ownership expectations generate small and insignificant shifts in valuation. These findings suggest that valuation is primarily anchored in the status quo of seed already held or routinely used, rather than in forward-looking expectations about prices or ownership. Under prevailing seed market conditions, such anchoring helps explain continued reliance on home-saved seed.

In Chapter 5, gender roles and social opportunity in shaping risk perceptions and adoption constraints are analyzed qualitatively. Using participatory qualitative methods, the chapter shows that men and women identify overlapping risks but rank them differently in line with socially assigned roles and responsibilities. Women place greater emphasis on production and health risks and face more limited access to formal seed and information channels, while men attach relatively greater importance to financial constraints. These differences imply that adoption barriers are not gender neutral. Formal seed sector interventions that focus narrowly on yield gains risk overlooking how legitimacy, household decision-making authority, and perceived household responsibilities shape women’s seed choices.

In Chapter 6, the effects of agricultural risk and the salience of background risk on investment in hybrid maize seed are assessed. Using a framed investment experiment with several decision rounds, the chapter tests how farmers adjust seed investment in response to production risk and background shocks. The results show that farmers systematically reduce investment as objective weather risk increases, consistent with reflective risk evaluation. Introducing an actuarially neutral background risk has little average effect on investment. However, framing background risk as a potential health shock selectively reduces women’s investment, while leaving men’s behavior largely unchanged. The gender-differentiated response does not appear to reflect systematic differences in baseline risk aversion, but rather a greater salience of health risks for women.

Chapter 7 serves as the synthesis chapter and advances the central integrative insight of the thesis: adoption outcomes reflect interactions between reflective processes responding to objective incentives and automatic responses shaped by experience, emotion, and social norms. While farmers adjust seed investment in response to explicit production risk in ways consistent with expected-utility logic, decision-making diverges when affect-rich narratives or habitual defaults become salient. In weakly credible seed markets, farmers often rely on habits, peer recommendations, and socially legitimate choices as substitutes for contested or unverified formal quality signals. Gendered social norms and household roles act as behavioral filters that legitimate action and shape automatic motivation.

This thesis contributes to development economics by linking behavioral mechanisms to smallholder seed investment under risk and uncertainty in Uganda’s maize seed market. Rather than attributing low adoption to a single binding constraint, the chapters highlight how information, reference points, and risk are interpreted in light of credibility, experience, and social norms. The synthesis indicates not an absence of demand, but a misalignment between seed products and the conditions under which smallholders evaluate and justify investment.