This article explores the developmental and geographical origins of economic preferences. For this purpose, we present a new data set that includes incentivized preference measures and survey data from more than 6000 individuals aged from 15 to 30 years. We link these data to respondents’ medical birth records, complete medical histories and to a wide variety of administrative records containing respondents’ own and their parents’ and siblings’ educational and labor market histories. Using machine learning methods, we explore the relationships between childhood circumstances, demographic variables, personality traits and economic preferences. The data reveal substantial heterogeneity in economic preferences across individuals and geographical regions. At the individual level, the data reveal robust correlations, esp. between birth weight and competitiveness and exposure to maternal smoking in utero and time preferences. We find suggestive evidence that children’s early health endowment and parental inputs act as complements, exacerbating the existing disparities in childhood endowments.