Granular Treasury Demand with Arbitrageurs

The U.S. Treasury market is the benchmark fixed-income market in the world, but it is not held by a single representative investor. Money market funds, banks, insurance companies, pension funds, mutual funds, foreign official institutions, foreign private investors, hedge funds, broker-dealers, and the Federal Reserve all hold Treasuries for different reasons. This paper asks how those different investors shape Treasury yields and the transmission of monetary policy.

The first contribution is data. We construct a granular dataset of U.S. Treasury holdings by investor sector and maturity bucket, covering most of the market from 2011Q4 to 2022Q4. This lets us ask who holds what part of the Treasury curve and how different investors respond to yields at different maturities.

The second contribution is to estimate maturity-specific demand functions. Investors do not only decide how many Treasuries to hold. They decide which maturities to hold. A money market fund is naturally concentrated at the short end. Pension funds and insurers have stronger long-maturity demand. Foreign official and private investors behave differently from domestic institutions. The paper estimates own-maturity and cross-maturity elasticities, which measure how demand for one maturity responds to yields at that maturity and at other maturities.

The third contribution is to embed these estimates into a dynamic equilibrium model with risk-averse arbitrageurs. In the model, arbitrageurs, empirically associated with broker-dealers and hedge funds, absorb imbalances across investor sectors and maturities. They are important because the Treasury market does not clear mechanically: when some investors sell long bonds or rebalance toward short bonds, someone must take the other side. Arbitrageurs do that, but only if compensated for bearing risk.

The paper's first main finding is that Treasury market elasticity is strongly maturity-dependent. The short end is more elastic because arbitrage is less risky there. The long end is less elastic because absorbing long-duration risk is costly. This means that demand shocks to long-term Treasuries can have much larger price effects than comparable shocks to bills.

The second finding concerns monetary policy. When the Federal Reserve tightens by raising short rates, some investors rebalance toward higher-yielding short-term Treasuries and away from long-term Treasuries. Arbitrageurs then must absorb more long-duration risk, which raises term premia. This helps explain why long-term rates can respond strongly to monetary policy shocks. Cross-maturity substitution is central to this mechanism.

The third finding concerns quantitative easing. Fed purchases have limited effects if investors expect them to be temporary. They matter much more if the Fed credibly commits to a sustained balance sheet expansion. This distinction is important for both QE and quantitative tightening. The market response depends not only on the size of the purchase today but also on beliefs about how persistent the Fed's demand will be.

The takeaway is that "the Treasury market" is not one demand curve. It is a collection of investor clienteles linked by arbitrageurs with limited risk capacity. Monetary policy and debt issuance affect yields by changing which investors must hold which risks. A policy that looks large in aggregate may have limited effect if it does not alter persistent demand. A policy that changes the long-run allocation of duration risk can have much larger effects.

This research provides a framework for thinking about Treasury market fragility, QE, and the future of debt issuance. As the Treasury market grows, understanding who absorbs new supply is as important as knowing how much supply is issued.