Building on Goodfriend and McCallum (2007), the paper presents a theory of money and banking in which the interbank Treasury-Bill (TED) spread (and other spreads) reflect the valuation of government bond collateral services in banking. Government debt held by households defrays the cost of borrowing from banks to finance deposits; and government debt held by banks facilitates the production of interbank credit in the provision of transactions services on deposits. The economy-wide collateral services yield is determined in general equilibrium by integrating the household and bank demand for collateral in an otherwise standard representative agent new synthesis (new Keynesian) model. Banking model parameters are calibrated to match average interest rate spreads and other banking aggregates from 1971 to 2006 (pre-crisis). The calibrated banking system demand for government bonds relative to deposits is a reasonably well-behaved function of the TED spread throughout the sample period. According to the calibrated model, the secular variation in the supply of government bonds relative to GDP over the sample explains the secular variation in the TED spread during the period. Significant short-term fluctuations in the TED spread are attributed to shocks to banking productivity. Spikes in the TED spread are purged of the collateral supply effects to measure and compare underlying distress in various banking crises during the sample period. The model is employed to assess the quantitative impact on various interest rate spreads of three policy exercises: an increase in the supply of government debt, an increase in aggregate bank reserves, and an exchange of government debt for bank reserves. The positive observed TED spread yields a calibrated bank loan to collateral ratio in excess of unity. Thus, the model predicts that a central bank open market purchase of debt for bank reserves creates a net scarcity of collateral, which raises the collateral services yield and elevates TED and other rate spreads.
