I study how quantities of safe bonds affect asset prices and lending volumes in financial markets. In a quantitative model, heterogeneous agents trade securities of different maturity and risk exposure. Risk-tolerant investors issue collateralized bonds to obtain leverage and to insure the risk-averse. Despite the presence of higher return assets, the most risk-tolerant hold long-maturity safe assets, which they value as good collateral. The value of collateralizability is high when safe bond quantities are low. Given measured variations in safe bond quantities between 1990 and 2015, the model replicates the dynamics of lending volumes and generates large, volatile credit spreads and excess return predictability. The model also predicts price effects of high-frequency changes of government debt quantities around tax due dates. In policy experiments, I use the model to study the effects of central bank asset purchases.