| This dissertation is composed of three chapters. In the first chapter, I show that the U.S. Treasury influences the yield curve by managing the auction cycles of its securities. First, I develop a simple theoretical model in which the Treasury selects the size and the frequency of auctions. I show that the Treasury may increase its revenues by choosing to issue small amounts frequently or larger amounts less frequently, and the relative profitability of each policy depends on its financing needs. Second, I estimate an empirical model of Treasury yields. I find that auction size and frequency affect new and old note yields as predicted by my model; in addition, their influence persists for several weeks after the initial issuance. Third, I evaluate the Treasury's decisions to shift the frequency of auctions between monthly and quarterly. Given the evolution of borrowing needs, I find that these policies saved U.S. taxpayers a considerable amount of money.;In the second chapter, I investigate demand effects in the primary market for Treasury securities. Recent empirical evidence shows that Treasury auctions exercise temporary pressure on bond prices. I study how the magnitude of this price pressure is related to the composition of buyers at auction. I distinguish between three main categories of investors: broker-dealers, investment funds, and foreign investors. I find that when dealers acquire a larger share of securities, auction yields tend to be lower and future excess returns higher. The opposite is true for investment funds and foreign investors. These findings are consistent with the view that broker-dealers typically act as "liquidity providers" of the Treasury market. They commit to buy supply in excess at auctions, but they require higher returns to be compensated for higher risk exposure.;In the third chapter, I examine the information content of bond spreads. Newly issued Treasury securities, called "on-the-runs", regularly trade at a premium over similar but older securities, called "off-the-runs". This spread is usually attributed to the different liquidity of the assets. By studying a simple portfolio choice model, I argue that movements in this premium also reflect arbitrageurs' risk-bearing capacity. Consistent with my hypothesis, I find empirically that the five- and ten-year on-the-run premia are related to a set of financial variables, to the current term structure, and to future excess returns. In addition, I document that these premia are positively related to primary dealers' positions. This evidence indicates that on-the-run premia may be a useful indicator of investors' risk exposure. |
