This paper reports the results of behavioral economic experiments conducted in Peru to examine the relationship amongst risk preferences, loan take-up, and insurance purchase decisions. This area-based yield insurance can help reduce people's vulnerability to large scale covariate shocks, and can also lower the loan default probability under extreme negative covariate shocks. In a context of collateralized formal credit markets, we provide suggestive evidence that insurance may help reduce the fear of losing collateral that prevents potential borrowers from taking loans. Framing these experiments to recreate a real life situation, we started with a Baseline Game where subjects had to choose between a fallback production project and an uninsured loan. We then introduced a third project choice--loan with yield insurance (Insurance Game)--which allows us to measure the effect of introducing insurance on the demand for loans. Overall, more than 50 percent of the subjects are willing to buy insurance in this insurance game. Further, controling for the number of peers in the ag network, wealth, and choices made in the baseline game, we find that the project choice decision is predicted by a judgment bias known as hot-hand effect, and risk aversion. In the latter case, the shape of the relationship is quadratic, meaning that highly risk averse subjects will prefer switching to the risky, uninsured loan project, while those showing a low and moderate risk aversion will stick to the safer (fallback or insured loan) projects.