An experimental test of whether financial incentives constitute undue inducement in decision-making

Many countries regulate the level of financial incentives offered for participation in activities such as clinical trials, egg donation, surrogacy, and organ donation. While ethical considerations surrounding exploitation and distributive justice play a role, a significant justification for these regulations is the concept of "undue inducement." This concept posits that offering substantial incentives can compromise an individual's rational decision-making process by excessively focusing their attention on the reward and potentially overlooking risks or negative consequences. The undue inducement hypothesis (UIH) comprises two parts: a positive claim, which predicts that incentives lead to biased information processing, and a normative claim, which asserts that this bias causes harm, making high-incentive transactions unacceptable even when low-incentive transactions are permissible. This paper directly investigates the cognitive underpinnings of the UIH through controlled experiments, addressing a significant gap in the existing literature, which primarily relies on small-scale case studies and unincentivized surveys. The research is vital because restrictive regulations, if not empirically justified, can hinder medical progress, limit options for infertile couples, and contribute to avoidable deaths. While undue inducement may not affect perfectly rational decision-makers, evidence suggests that real-world decision-making is often suboptimal. The experiments use incentivized activities with downsides commensurate to the offered payments, ensuring the incentives are significant but not so large as to override rational judgment.

The existing literature on undue inducement primarily consists of case studies and unincentivized surveys, yielding suggestive rather than conclusive evidence. While some studies suggest payments don't significantly affect judgments about risks in research participation, they lack the rigor of real choices and formal welfare analysis. Research on motivated reasoning, which often focuses on situations where bias benefits the decision maker, provides limited insights into the UIH, where bias is hypothesized to be harmful. This study aims to fill this gap by conducting controlled experiments with substantial stakes, where participants' actual decisions are observed. By analyzing real choices and formally assessing welfare consequences, it directly tests the core claims of the UIH, offering a more robust assessment of the impact of incentives on decision-making.

The paper employs two main experiments. Experiment 1 (n=671 US undergraduates) uses a visceral and aversive transaction: eating insects. Participants were randomly assigned to receive either $3 or $30 for consuming insects. Before making their decision, they could choose between videos highlighting the pros and cons of eating insects. The experiment tests UIH-positive by examining if higher incentives lead to a greater preference for the encouraging video, and UIH-normative by comparing the ex-post surplus (difference between the incentive and the participant's reservation price, representing their welfare) between the two incentive groups. To address concerns about anchoring effects and video bias, the analysis uses a counterfactual surplus calculation based on reservation prices from a control group without video access. Experiment 2 (n=406 German undergraduates) involves a riskier financial decision: risking a €100 loss for a payment of €20-€80. Participants choose between "bold" and "cautious" advisors providing imperfect recommendations before deciding. UIH-positive is tested by observing whether higher incentives increase the preference for the bold advisor. UIH-normative is tested using the reframed decisions paradigm, comparing the participants' choices to their certainty equivalents for equivalent lotteries. This approach avoids hindsight bias by assessing decisions based on information available at the moment of choice. In both experiments, the welfare analysis considers different social welfare functions and uses various statistical techniques to account for noise and potential biases in data. A theoretical model of costly information acquisition clarifies the relationship between UIH-positive and UIH-normative, demonstrating that the former doesn’t necessarily imply the latter.

Experiment 1 confirmed UIH-positive: higher incentives increased the demand for the encouraging video and biased information acquisition toward positive information. However, UIH-normative was violated. While a substantial minority (10-20%) of participants made bad decisions, increasing incentives did not exacerbate these mistakes. Welfare analysis consistently favored the higher incentive regardless of the social welfare function considered, even after addressing potential anchoring effects and video bias. Experiment 2 also supported UIH-positive: higher incentives increased the preference for the bold advisor. Again, UIH-normative was violated; mistakes were common, but their frequency did not significantly change with incentive amounts. This held true across treatments varying stakes, delay of consequences, and risk probabilities. The results suggest that even suboptimal decision-making is not significantly amplified by higher incentives. Analysis demonstrated that both information biases observed (supporting UIH-positive) were consistent with Bayesian rationality; this is because rational actors, facing costly information search, will favor sources that minimize the costs of more expensive errors.

The experiments provide robust evidence that while higher incentives lead to biased information search (consistent with UIH-positive), this bias doesn't necessarily translate into lower welfare (contradicting UIH-normative). A theoretical model explains this: rational agents, facing limited information acquisition, strategically select sources that reduce the cost of the more expensive type of error. This finding has significant implications for policy. Given the potentially high societal costs of unduly restricting voluntary transactions based on the UIH, the paper suggests that regulations limiting incentives based on undue inducement concerns should be reconsidered. This conclusion contrasts with the normative aspects of the UIH, suggesting a need for more nuanced approaches to incentivized transactions.

This research demonstrates that financial incentives, while impacting information-seeking behavior, do not systematically lead to worse welfare outcomes in the contexts studied. This challenges the long-held assumption underpinning regulations limiting incentives based on undue inducement. Future research should extend these findings to other contexts, participant populations, and transactions currently subject to such regulations, while considering potentially different psychological mechanisms and incorporating broader ethical considerations beyond welfarism.

The experiments, though rigorously designed, are limited to laboratory settings. While the tasks were chosen to have high ecological validity, real-world transactions may involve additional complexities not fully captured. The participant pools (US and German undergraduates) may not fully represent the diversity of populations involved in the types of transactions studied. The findings might not fully generalize to settings with highly vulnerable populations or those involving transactions with more significant and irreversible long-term consequences.