The importance of risk-sharing in agricultural economies has been extensively analyzed through the principal-agent framework, which predicts that sharecropping should be observed more frequently than fixed rent contracts when output uncertainty is higher. Empirical studies, however, provide mixed support for this prediction, since often fixed rent contracts are found to be prevalent in more risky environments. The first chapter provides a model where the relative incidence of share tenancy over fixed rent contracts may be negative depending, among other things, on the relative average degree of risk aversion of tenants and landlords. The second chapter explores the empirical validity of the theoretical framework using Indian data. After paying special attention to the measure of uncertainty used to identify farming risk, a parameterized version of the theoretical model is structurally estimated. The econometric results support the proposed model. The third chapter studies the offset effect of pension wealth on private wealth when individuals are misinformed about their future retirement benefits. We show that if individuals have expectational errors correlated with their actual pension wealth, and update over time their expectations, then the canonical econometric specification used so far to estimate the offset effect gives biased estimates. An alternative econometric specification is proposed and used to estimate the offset effect on Italian data. The estimates obtained are higher than the ones previously found in the literature.