When the Fed Shrinks, the Repo Constraint Moves

This chapter examines the balance-sheet mechanism that connects a reduction in the Federal Reserve’s Treasury holdings to repo funding demand, MMF lending choices, and bank reserves. The point is not to memorize a particular T-account. It is to learn how the same balance-sheet action can tighten repo capacity while producing different reserve movements depending on the amount of cash held in the ON RRP facility.

The starting point is an asset reallocation. When the Federal Reserve reduces its holdings of Treasury securities, those securities do not disappear from the financial system. More of them must be held by private investors. Primary dealers may hold a share of the additional inventory, and they commonly finance Treasury positions by borrowing in repo. A smaller Fed balance sheet can therefore mean a larger private-sector financing task.

That statement can sound paradoxical if “shrinkage” is heard only as a reduction in liquidity. The Fed’s action changes at least two linked quantities: the distribution of Treasury securities and the location of cash used to finance them. The first change tends to increase repo demand. The second determines which institutions absorb the cash movement and how much flexibility remains on the lending side. Separating those movements is the key to understanding why reserves and repo conditions need not move in lockstep.

The asset-side movement: more Treasuries to finance

Imagine a simplified balance sheet before the Federal Reserve reduces its Treasury holdings. The Fed owns a portion of the Treasury stock, while the private sector—including dealers—owns the rest. The Fed also supplies liquidity to the private sector through bank reserve balances and, when used, ON RRP balances held by nonbanks. If the Fed’s Treasury assets decline, the private sector must hold a larger share of the outstanding Treasury securities.

A dealer is one of the private-sector institutions that can absorb that change. Dealers purchase Treasury securities in the primary market and make markets in them. If a dealer holds more securities than it wants to fund from its own cash, it can borrow overnight in repo against the Treasury collateral. The immediate accounting logic is straightforward: more private Treasury inventory creates a larger request for financing; the market consequence depends on whether repo lenders can meet that request without demanding a higher rate.

The important word is “can.” Balance-sheet reduction does not force every additional Treasury security into a dealer’s inventory, and it does not mean every private holder finances the security through the same repo segment. The Federal Reserve’s example is explicitly stylized. It establishes the direction of the channel—less Fed-held Treasury stock, more private holdings, and greater dealer demand for repo funding—without identifying a universal amount of inventory that must be financed or a fixed rate response.

A useful way to follow the T-account logic is to keep the two sides separate. On the Fed’s asset side, Treasury holdings become smaller. On the private side, Treasury holdings become larger. The corresponding claims and payments are distributed across the institutions involved; what changes for the market is who must carry the securities and arrange their financing. The operation is therefore not equivalent to money vanishing. It is a reallocation that leaves a larger portion of the Treasury stock in hands that may rely on short-term funding.

The dealer then has a choice at the next margin. It can use its own cash, find a buyer for some of the securities, or borrow against them in repo. The source’s mechanism focuses on the last option because it directly connects private Treasury holdings to the demand for repo funding. But the existence of alternatives is part of the explanation too. If the dealer can distribute the securities easily, the extra repo demand may be limited. If many dealers are trying to fund inventory at once, or if selling would be costly, more of them may approach repo lenders simultaneously. The rate is shaped by the interaction of those choices, not by the asset transfer alone.

The collateral itself is not the scarce object in this story. Treasury securities can be high-quality and widely accepted while the balance-sheet space used to carry them is limited. A dealer that is comfortable holding the securities still needs a cash lender. If funding is readily available, the larger private position may be absorbed with little visible rate pressure. If the marginal lender is reluctant or already committed to another use of cash, the dealer may have to pay more, sell inventory, or reduce the position. Repo capacity is the ability of the system to perform that financing task, not simply the existence of acceptable collateral.

This also clarifies why the effect of balance-sheet policy is not captured by looking only at the Fed’s asset total. Two balance sheets with the same broad size can have different implications if one leaves a different mix of Treasury bills and coupon securities with the private sector, or if the cash supporting MMF lending is held in a different facility. Asset composition and liability composition affect who must act at the margin.

The mechanism can be summarized as a sequence with a fork on the cash side:

The fork should not be read as two different asset-side mechanisms. Dealer demand for repo rises from the same private-sector Treasury reallocation in both branches. What changes is the source from which an MMF can obtain or redeploy cash when it wants to lend more.

The cash-side fork: ON RRP versus bank deposits

ON RRP balances are important because they provide a nonbank cash buffer at the Federal Reserve. When an MMF has money in ON RRP and repo demand becomes stronger, it can substitute those ON RRP balances for repo lending. In the Federal Reserve’s stylized example, the MMF’s cash moves from one claim on the Fed to a private repo transaction. Reserves and bank balance sheets are unchanged, even though repo capacity is being used to finance more private Treasury holdings.

This case shows why a rise in repo lending does not necessarily require a fall in bank reserves. The marginal cash has already been held by a nonbank at the Fed. The MMF can release that cash into repo without first drawing deposits away from a bank. From the perspective of the banking system, reserve balances have not become the adjustment variable. From the perspective of dealers, however, the market still has to decide whether the additional funding is available and at what price.

There is a subtle but important distinction here between a change in the location of cash and a creation of new cash. The MMF does not need a new reserve injection to make the substitution. It changes a position at the Fed into a private lending position. That can relieve the immediate pressure on the dealer while leaving the banking system’s reserve balances unchanged. The nonbank buffer has done the absorbing. If that buffer is large enough and lenders are willing to use it, a balance-sheet change can be visible in repo activity without yet appearing as a reserve shortage.

The situation changes when ON RRP take-up is close to zero. There is little or no nonbank balance at the facility for MMFs to convert into repo lending. If MMFs want to provide more cash, they must attract new deposits away from banks. As deposits leave the banking system, bank reserve holdings decline. Here the repo market’s financing demand and the banking system’s reserve movement become linked through the MMF’s funding search.

The two cases can be set beside each other:

This table is not a claim that the two states exhaust every institutional possibility. It isolates the contrast the Federal Reserve note uses to explain why the same balance-sheet change can show up differently in reserves. Other repo segments and other private responses may matter, but they do not erase the central distinction between drawing down an existing nonbank balance and competing for bank-linked cash.

The comparison also changes how to interpret the phrase “reserve drain.” In the near-zero ON RRP case, the reserve decline is part of the mechanism by which MMFs obtain the cash needed to lend. In the positive-ON RRP case, repo lending can expand while bank reserves remain unchanged because the source of cash is already outside the banking system. Calling both outcomes a reserve drain would hide the institutional difference. Calling neither outcome a liquidity event would also be misleading: the market-facing supply of repo cash and the location of the cash have changed.

The sum of reserves and ON RRP is useful precisely because it captures aggregate Fed-provided liquidity to the private sector. Reserves are cash held at the Fed by banks; ON RRP balances are cash held there by nonbanks. Adding them gives a system-wide measure of the liquidity supplied through those channels. But an aggregate is not a deployment schedule. It does not say which institution holds the cash, which portfolio choice is most attractive, or how much will be offered in repo on a particular day.

That is the difference between a reservoir and a pipeline. A reservoir can contain a large volume while the pipeline connecting it to a particular borrower is narrow or temporarily unused. In this context, the relevant pipeline is made of dealer balance sheets, MMF portfolio decisions, and the market arrangements that match Treasury collateral with cash. The Federal Reserve’s daily estimates later in the analysis find that changes in reserves plus ON RRP do not directly explain daily repo spreads in their specification. The note interprets that result as consistent with dealer and MMF behavior determining how much aggregate liquidity is actually deployed into repo.

This does not make aggregate liquidity irrelevant. It can determine how much room exists before an adjustment must draw on banks, and it can affect the state in which a Treasury-supply shock arrives. A positive ON RRP balance can absorb some reallocation without immediately reducing reserves. Near-zero ON RRP removes that buffer. The quantity matters through the institution that holds it, the alternative uses available to that institution, and the sequence in which borrowers and lenders respond.

The same logic helps explain why the repo constraint can appear before bank reserve demand becomes binding. Dealers may face a larger financing requirement as more Treasury securities sit with the private sector, while banks still hold reserves above the level at which their own federal-funds behavior becomes sharply sensitive. Repo rates can rise first because the marginal nonbank lender, not the marginal bank borrower, is the point of stress. Only if conditions spill over—or if MMFs must attract deposits—does the connection to bank reserves become more visible.

Repo and federal funds remain distinct markets. The Federal Home Loan Banks are major federal-funds lenders and also operate in repo. Their different market choices can transmit repo pressure into EFFR, even though participants and trading times differ. That transmission is an additional step, not part of the accounting identity itself. The balance-sheet mechanism tells us where pressure originates; the cross-market choice tells us how far it can travel.

The result is a more precise interpretation of balance-sheet reduction. It does not simply remove a quantity called liquidity. It reallocates Treasury holdings toward the private sector, raises a potential demand for dealer repo funding, and changes whether MMFs can meet that demand from ON RRP balances or must compete for bank deposits. To determine how strong the resulting rate response is, the next question is empirical: how does repo sensitivity change as aggregate liquidity becomes lower relative to the size of the economy?

References

• Repo Markets and the Fed’s Balance Sheet: Implications for Monetary Policy Implementation