Intermediary Balance Sheets and the Treasury Yield Curve

U.S. Treasury securities are usually described as the safest and most liquid assets in the world. That is true, but it is incomplete. Treasuries still have to be held, financed, and intermediated by financial institutions. When the balance sheets of those institutions become costly, Treasuries can trade as if they are inconvenient to hold, even though they remain safe and liquid.

This paper studies a striking regime change in the Treasury market after the Global Financial Crisis. Before the crisis, primary dealers were typically net short Treasury bonds, and Treasury yields were below comparable swap rates. After the crisis, dealers became net long Treasuries, and Treasury yields moved above comparable swap rates. In market language, swap spreads turned negative. That change was surprising because the traditional view was that Treasuries should trade at lower yields because of their convenience.

The key idea is that convenience is not only about the asset's intrinsic safety. It also depends on who must hold the asset and how costly it is for them to finance it. Dealers intermediate Treasury markets. If the rest of the market wants to hold fewer Treasuries than the government supplies, dealers absorb the extra supply and become net long. Holding those positions uses balance sheet capacity. When balance sheet capacity is expensive, dealers require compensation, and Treasury yields can rise relative to swap rates.

We construct two benchmark yield curves. The "net-short" curve measures the yield level at which a dealer would be willing to be short Treasuries after accounting for financing and balance sheet costs. The "net-long" curve measures the yield level at which a dealer would be willing to hold Treasuries. Actual Treasury yields tracked the net-short curve before the crisis and the net-long curve after the crisis. This pattern lines up with the observed shift in dealer positions.

The paper also connects Treasury pricing to covered interest parity deviations, another post-crisis phenomenon. Covered interest parity is a basic no-arbitrage relation in global funding markets. Persistent deviations from it suggest that intermediary balance sheets are scarce. We show that the same balance sheet costs that help explain currency-market arbitrage violations also help explain negative Treasury swap spreads. These are not separate puzzles; they are related symptoms of balance-sheet-constrained intermediation.

The policy implication is that the effect of monetary and regulatory policies depends on the Treasury-market regime. A policy that expands Treasury supply or changes the balance sheet cost of holding Treasuries may have different effects when dealers are net long than when they are net short. This helps explain why Treasury yields can behave differently across crises. In the Global Financial Crisis, Treasuries became more valuable relative to other rates. In March 2020, stress in Treasury intermediation pushed in the opposite direction.

The main message is that the Treasury market is not frictionless plumbing. It is a market that depends on the capacity of intermediaries. As government debt grows, the private sector must absorb more Treasuries. If that absorption falls on balance-sheet-constrained dealers, Treasury securities can become expensive to intermediate even while remaining safe to own. That changes how debt issuance, regulation, and central bank interventions transmit through the financial system.

This distinction is important for current debates about government debt. The question is not only whether the United States can issue more safe debt. It is also who will hold that debt, on what balance sheet, and at what cost. The answer can affect the entire yield curve.