Intertemporal choice experiments are increasingly implemented to make inference about discounting and marginal utility, yet little is known about the predictive power of resulting measures. This project links standard experimental choices to a consumption smoothing decision with large stakes — around 10% of annual income. In a sample of around 400 Guatemalan Conditional Cash Transfer recipients, we find that preferences over large-stakes payment plans are significantly correlated with experimental measures of patience and diminishing marginal utility. These represent the first findings in the literature on the predictive content of such experimentally elicited measures for a large-stakes decision.