This paper studies how a financial system that is organized to efficiently create safe assets responds to macroeconomic shocks. Financial intermediaries face a cost of bearing risk, so they choose the least risky portfolio that backs their issuance of riskless deposits: a diversified pool of nonfinancial firms' debt. Nonfinancial firms choose their capital structure to exploit the resulting segmentation between debt and equity markets. Increased safe asset demand yields larger and riskier intermediaries and more levered firms. Quantitative easing reduces the size and riskiness of intermediaries and can decrease firm leverage, despite reducing borrowing costs at the zero lower bound.

We develop a general equilibrium model in which households’ mortgage leverage is determined by supply and demand forces, where the price of credit impacts the quantity of leverage households choose. Mortgages are supplied by financial intermediaries, who offer households a menu of mortgage contracts whose pricing varies with intermediaries’ equity capital. In the model, growth in the demand for safe assets that replicates the falling interest rates in the 2000s causes an empirically realistic boom in household borrowing, debt-financed consumption, and house prices. This boom results in a larger bust in asset prices and household borrowing in future financial crises.

We infer risk‐free rates from index option prices to estimate safe asset convenience yields in 10 G11 currencies. Countries' convenience yields increase with the level of their interest rates, with U.S. convenience yields fifth largest. During financial crises, convenience yields grow, but the difference between United States and foreign convenience yields generally does not. Covered interest parity (CIP) deviations using our option‐implied rates are a similar size between the United States and each other country. A model in which convenience yields depend on domestic financial intermediaries, but CIP deviations reflect the funding costs of international arbitrageurs financed with dollar‐denominated debt, explains these results.

We analyze interactions between fiscal and monetary stimulus in a new Keynesian model with nominal mortgage debt that can be inflated away. Redistributive transfers are most impactful when followed by a temporary deviation from inflation-targeting monetary policy. Unlike other fiscal policies, redistribution causes inflation even in the absence of long-run debt sustainability problems, and inflating away mortgages results in additional redistribution. In a quantitative model with mortgage refinancing frictions, transfer payments provide much more stimulus than the redistributive effects of inflation. Borrowers gradually refinance after their mortgages are inflated away, dampening their immediate consumption response and reducing their long-run labor supply.

We estimate risk-free interest rates unaffected by convenience yields on safe assets. We infer them from risky asset prices without relying on any specific model of risk. We obtain interest rates and implied convenience yields with maturities up to three years at a minutely frequency. Our estimated convenience yield on Treasuries equals about 40 basis points, is larger below three months maturity, and quadruples during the financial crisis. In high-frequency event studies, conventional and unconventional monetary stimulus reduces our rates more than the corresponding Treasury yields, thus broadly affecting rates even outside the narrow confines of the fixed-income market.

We find that central bank reserves injected by QE crowd out bank lending. We estimate a structural model with cross-sectional instrumental variables for deposit and loan demand. Our results are determined by the elasticity of loan demand and the impact of reserve holdings on the cost of supplying loans. The reserves injected by QE raise loan rates by 7.4 basis points, and each dollar of reserves reduces bank lending by 7.7 cents. Our results imply that a large injection of central bank reserves has the unintended consequence of crowding out bank loans because of bank balance sheet costs.