This paper studies differences in production structures across countries and their implications for crosscountry heterogeneity in GDP volatility. In particular, economies with more input-output connections-a denser network-are associated with less concentrated sales shares and lower volatility. The relationship between density and volatility is stronger in countries with a higher share of services in GDP (hereafter referred to as service share). To account for this evidence, I propose a generalized production network model in which denser economies display higher production complexity. If production is also specialized in industries that use labor and intermediates as substitute inputs, higher network density indeed lowers the concentration of sales shares and aggregate volatility. U.S. sectoral data suggest that the elasticity of substitution between labor and intermediates in service sectors is larger than one and larger than in non-service sectors. A calibrated model that then also matches each country's production network can quantitatively generate observed crosscountry empirical patterns. Furthermore, in contrast to previous work, the model predicts that: i) sectoral shocks play only a modest role in accounting for the observed business cycle dynamics, especially in dense and service-oriented economies; and ii) production diversification does not always lower volatility.

Cross-sectoral heterogeneity in sectoral bond spreads is related to sectoral elasticities of substitution in production. During the Great Recession, more flexible firms paid lower sectoral bond spreads, generated higher revenues, and held more working capital. A model consistent with these facts-input-output linkages, working capital constraints, and heterogeneous elasticities-predicts that sectoral distortions during the Great Recession generated an efficiency wedge-due to input misallocation-2.4 times larger than one with homogeneous production functions. In addition, our model predicts input-output connections amplified the Great Recession 2.3 times as much as one with homogeneous elasticities.

We develop a theory of saving-constrained households to explain the following facts that are difficult to reconcile with existing theories: 1) Consumption is excessively volatile relative to income (established fact), 2) a large fraction of high-debt households exhibit marginal propensities to consume near zero, 3) lagged high expenditure is associated with low contemporaneous spending propensities. Our proposed interpretation of these facts is that household expenditure depends on time-varying minimum consumption thresholds that, if violated, yield substantial utility costs. We demonstrate that such a model can match many features of the joint dynamics of income and consumption. Our theory has implications for the propagation of macroeconomic shocks.

We document four features of consumption and income microdata: (1) householdlevel consumption is as volatile as household income on average, (2) householdlevel consumption has a positive but small correlation with income, (3) many low-wealth households have marginal propensities to consume near zero, and (4) lagged high expenditure is associated with low contemporaneous spending propensities. Our interpretation is that household expenditure depends on timevarying consumption thresholds where marginal utility discontinuously increases. Our model with consumption thresholds matches the four facts better than does a standard model. Poor households in our model also exhibit "excess sensitivity" to anticipated income declines.

I study optimal sectoral policies in models with input-output linkages and distortions. Labor subsidies cannot implement the first best allocation. Intermediate input subsidies, or the right combination(s) between subsidies to labor and intermediates, can optimally correct for network externalities.