Research

This paper studies how people use control in relationships based on trust. In many real-world interactions—from households to informal markets—partners seek to shape one another’s behavior through threats, sanctions, or restrictions—whether a spouse withholding shared resources or a lender threatening to seize a borrower’s collateral—yet such controlling behavior is difficult to observe in standard data. To address this, I design a modified trust game in which trustors choose not between trust and distrust, but between trusting freely and attempting to control—by pre-committing to cut total payoffs in half if the trustee acts unfairly. This instrument introduces a destructive but decentralized form of deterrence, capturing how informal enforcement operates when formal institutions are weak. I implement a $2 \times 2$ lab-in-the-field experiment in Canada and Peru, cross-randomizing the availability of the control instrument and participant location. Results show that control consistently increases reciprocity in both settings, but only raises trust in Peru, where formal enforcement is weaker. Moreover, trustors who refrain from controlling—even when the option is available—earn higher returns, suggesting that restraint itself functions as a credible signal of goodwill. Together, these findings reveal the dual role of informal enforcement: instruments of control can both sustain and corrode cooperation, depending on the institutional context.

In this paper we report on a lab experiment in the field in which we measure differences in beliefs about gender norms between Peruvian men and women, and how these differences vary according to different reference groups against which individuals hold common beliefs. Specifically, we ask whether common beliefs vary according to whether the reference group in consideration is peer men or peer women. We find that men and women do not have vastly different private beliefs about gender equitable norms. However, both groups understate their peers' inclination for gender equity. This is especially the case for women who believe that peer men are drastically more disinclined towards gender equity than men actually are.  These results are largely driven by more educated participants who are significantly more likely to misperceive men's beliefs. We verify our findings against the possibility that social desirability bias influences how participants respond to the belief elicitation instruments.  We find evidence that men exhibit significantly more socially desirable behaviour in their responses, but the asymmetric misperceptions along gender and education dimensions is present among participants with low and high social desirability scores.  Our results suggest that policy aiming to correct gender norm beliefs misperceptions may need to customize the informational content of their efforts to account for relevant reference groups and possible socio-economic characteristics of individuals who hold such misperceptions.

This paper develops a non-cooperative model of intrahousehold bargaining to analyze how spousal discordance—defined as a mismatch between expectations and experiences in household decision-making—shapes autonomy, conflict, and punishment dynamics. I model a sequential game in which the husband can exert an ex-ante threat of shading: a strategic form of non-cooperation that includes subtle acts such as emotional withdrawal, implicit disapproval, or undermining autonomy without overt violence. The framework relaxes the assumption of common knowledge over bargaining power and treats discordance as both a source of inefficiency and a trigger for punitive behavior. The model yields several novel insights: shading is not necessarily decreasing in the wife’s reservation utility, nor necessarily increasing in the husband’s. The threshold required to force the wife into a low-empowerment state depends jointly on her empowerment probabilities, her disutility from punishment, the distance between empowerment types, and household preferences over public goods. From the husband’s perspective, the incentive to refrain from shading increases with his own disutility from conflict, the cost of discordance, and expectations about the wife’s likely response. These results highlight that sustainable female empowerment requires shifting not only women’s outside options, but also the opportunity cost of coercion for men—suggesting a dual-lever policy approach.

This paper examines how women’s intra-household bargaining power shapes modern contraceptive use in Zambia, using data from the 2018 Demographic and Health Survey (DHS). Bargaining power is proxied through standard DHS indicators of decision-making autonomy (e.g., healthcare, purchases, mobility, contraception) and financial capability. A core contribution is the construction of couple-level disagreement metrics: using matched partner data, I distinguish not only between concordance and discordance, but also the direction of mismatch—whether women claim more decision-making power than their husbands report (“power-taking”), or less (“power-giving”). Women with both financial autonomy and sole authority over healthcare decisions are 87% more likely to use modern contraception; sole authority over contraceptive decisions alone increases this probability by 56%. Couples with shared recognition of women’s empowerment show higher uptake (16.7% increase), as do couples where women assert more power than their husbands acknowledge (14.6%), consistent with feminist theories of “power within.” I also apply LASSO-based model selection methods (cross-validated LASSO, BIC minimization, and adaptive LASSO) to test robustness and predictive stability. While the compound empowerment model performs consistently across specifications (R2> 0.72), the discordance model underperforms—highlighting the empirical limits of modeling subjective household dynamics. To interpret these findings, I propose a stylized behavioral framework—the Compound-Discordance Model—which formalizes how empowerment, financial autonomy, and spousal alignment interact to shape contraceptive behavior. Together, the empirical and theoretical results underscore the need to move beyond static, additive models of “who decides,” toward a more relational and psychologically grounded account of reproductive autonomy.