We develop a non-linear, quantitative macroeconomic model with heterogeneous monopolistic financial intermediaries, incomplete markets, default risk, endogenous bank entry, and aggregate uncertainty. The model generates a bank net worth distribution fluctuation problem analogous to the canonical  Bewley-Huggett-Aiyagari-Imrohoglu environment. Our frameworknests Gertler and Kiyotaki (2010) (GK) and the standard Real Business Cycle model as special cases. We present four general results. First, relative to the GK benchmark, banks’ balance sheet-driven recessions can be significantly amplified, depending on the interaction of endogenous credit margins, the cyclicality of a precautionary lending motive and the (counter-) cyclicality of intermediaries’ idiosyncratic risk. Second, equilibrium responses to aggregate exogenous shocks depend explicitly on the conditional distributions of bank net worth and leverage, which are endogenous time-varying objects. Aggregate shocks to banks’ balance sheets that hit a concentrated and fragile banking distribution cause significantly larger recessions. A persistent consolidation in the U.S. banking sector that matches the one observed over 1980-2020 generates a large economic contraction and an increase in financial instability. Third, we document, and match, novel stylized facts on both the cross-section of credit margins and the cyclical properties of the first three moments of the cross-sectional distributions of financial intermediary assets, net worth, leverage, loan margins, and default risk. We find that shocks to capital quality and to leverage constraint tightness (“financial shocks”) can match fluctuations in the U.S. financial sector very well. Finally, we use the model to identify and characterize episodes of systemic banking crises. Such events are associated with large economic recessions, spikes in bank leverage, and large drops in the number of intermediaries.