Using a unique dataset covering the universe of Portuguese firms and their credit situation we show that financially constrained firms are found across the entire firm size distribution, even in the top 1%. Incorporating a richer, empirically supported, productivity process into a standard heterogeneous firms model generates a joint distribution of size and credit constraints in line with the data. The presence of large constrained firms in the economy, together with their elevated capital share, explains about 66% of the response of output to a financial shock. We conclude by providing micro-evidence in support of the model mechanism.

A simple model predicts that firms with lumpy investment profiles face elevated external finance premia, leading to a reduced propensity to adjust. This creates lumpy investment spirals - firms failing to adjust also face higher premia in the future, decreasing their propensity to invest tomorrow. Using Compustat data we show that these predictions are consistent with observed investment lumpiness and risk premium patterns. Finally, we analyse how the presence of lumpy investment spirals affects capital misallocation and aggregate shock sensitivity in a heterogeneous firms model with financial frictions.

This paper solves a reduced history model with heterogeneous agents, aggregate risk and portfolio choice between a risky, productive and a risk-free, non-productive asset. Motivated by existing empirical evidence and consistent with the portfolio choice theory of precautionary savings I find that the idiosyncrasy in human capital gives rise to heterogeneous asset allocations. The agents with the lower consumption growth save more in the risky asset, whilst poorer agents only use the safe asset, implying that the relative supply of safe assets is a central determinant for the evolution of aggregate consumption. Separately, this paper also improves upon the existing methodology of reduced history representations by finding a global solution, enabling studies of richer environments within this framework. In particular, a study of optimal fiscal policy in a heterogeneous agents model with portfolio choice becomes feasible.