*This paper examines the long-run effects of household income shocks on children’s outcomes with different family structures. In particular, we exploit exogenous variation in rainfall across districts in Tanzania to investigate the extent to which transitory income shocks lead to different consequences on education and marital status by gender, birth order and family size. I find that having brothers can reduce the negative impact of income variation on girl's education while the number of sisters in the household does not affect boy's schooling. Moreover, negative deviations in rainfall from the long-term mean impact positively older sisters’s marital status, marrying at a later age than their younger sisters while no significant effects are identified for boys. This study is consistent with a model in which parents respond to income shocks by allocating differently resources to girls especially the youngest, while boys are quite sheltered.

**We investigate the extent to which standard one sector RBC models with positive externalities and variable capacity utilization can account for the large hump-shaped response of output when the model is submitted to a pure sunspot shock. We refine the Benhabib and Wen (2004) model considering a general type of additive separable preferences and a general production function. We provide a detailed theoretical analysis of local stabilities and local bifurcations as a function of various structural parameters. We show that, when labor is infinitely elastic, local indeterminacy occurs through Flip and Hopf bifurcations for a large set of values for the elasticity of intertemporal substitution in consumption, the degree of increasing returns to scale and the elasticity of capital-labor substitution. Finally, we provide a detailed quantitative assessment of the model and conclude with mixed results. We show that although the model is able theoretically to generate a hump-shaped dynamics of output following an i.i.d. sunspot shock under realistic parameter values, the hump is too persistent for the model to be considered fully satisfactory from an empirical point of view.