How do people use their past learning experiences when deciding whether to invest in new learning opportunities? These experiences likely provide relevant information about learning abilities, which could be used to improve forecasts about future learning trajectories, leading to better choices in skill acquisition. However, we know little about how people actually use information from prior learning to guide their future skill investment decisions. To answer this question, I conduct a controlled experiment where participants complete a real-effort task that requires learning new information. By exogenously manipulating participants' learning curves, I study how beliefs about future learning trajectories and willingness to invest in future skill acquisition react to previous experiences. This design lets me compare participants' belief updating against Bayesian predictions and identify how cognitive biases might prevent accurate learning from experience. The results reveal systematic mispredictions about learning trajectories that persist and often worsen after gaining experience, affecting skill investment decisions. This research contributes to the literature on human capital formation and decision-making under uncertainty, with important implications for educational practices and workforce development strategies. 

Do discounts affect future demand through psychological mechanisms inconsistent with rational responses? We study this question through a novel experimental design able to detect both anchoring (whereby discounts decrease future willingness to pay) and reciprocity (whereby discounts increase future willingness to pay). In a field experiment in Malawi, we randomly assign discounts for preventive health care products. We document that past discounts 1) significantly reduce future willingness to pay - with clear evidence of anchoring on previously seen prices - but 2) also significantly decrease the share of individuals who declare that they would never buy that product. An additional experiment that randomizes non-price features of the transaction provides evidence that the latter is consistent with reciprocity. Design choices of our experiment allow us to rule out rational explanations for those patterns, such as intertemporal substitution, price-based inference about quality or learning about costs. Results have the potential to reconcile conflicting findings from the literature and to inform the optimal design of subsidies in preventive healthcare and beyond. 

Despite substantial advances in women's educational attainment, significant gender gaps persist in labor market outcomes. Emerging evidence suggests women make career-limiting choices based on perceptions of what potential partners value, yet we know little about whether people accurately perceive others' partner preferences. We conduct an experiment using a Swiss dating app with over 28,000 active users, implementing incentive-compatible methods to measure both actual partner preferences and second-order beliefs about others' preferences across seven key characteristics. We find substantial and systematic misperceptions, suggesting that women's career-limiting choices may be based on inaccurate perceptions of what men actually value, potentially perpetuating gender stereotypes and contributing to persistent labor market inequality. Building on these findings, we conduct a follow-up field experiment with speed dating events to measure how correcting these misperceptions affects marriage market outcomes as well as investment decisions in different skills. 

Coming soon